Waste and organized crime in regional environments

Resource and Energy Economics 41 (2015) 185–201

Contents lists available at ScienceDirect

Resource and Energy Economics journal homepage: www.elsevier.com/locate/ree

Waste and organized crime in regional environments How waste tariffs and the mafia affect waste management and disposal Alessio D’Amato a,b, Massimiliano Mazzanti c,b,∗, Francesco Nicolli d,b a b c d

CEIS and DEF, University of Rome “Tor Vergata”, Italy SEEDS, Italy University of Ferrara, Italy IRCRES CNR Milan, Italy

a r t i c l e

i n f o

Article history: Received 26 March 2012 Received in revised form 6 April 2015 Accepted 11 April 2015 Available online 22 April 2015 JEL classification: Q53 R11 K42 Keywords: Organized crime Waste management and disposal Enforcement costs Decentralized policy Panel data Municipal waste

a b s t r a c t Waste management and disposal are influenced by socio-economic, institutional and policy factors that possess idiosyncratic features in regional settings. The role of organized crime is a largely unexplored factor. Crime organizations such as the mafia are known to collude with local institutions to control waste markets. As a result, legal forms of waste disposal and socially preferable management options are often undermined primarily through an influence on policy enforcement. Given its high regional heterogeneity and local ‘waste crises’, Italy provides a compelling case for the study of crime’s effects on local waste performance. Panel econometric analyses show that sorted collection of recyclable waste and legal forms of waste disposal are lower when the mafia’s effect on the actions of local governments is more intense. © 2015 Elsevier B.V. All rights reserved.

∗ Corresponding author at: University of Ferrara, Italy. Tel.: +39 0532 455066. E-mail address: [email protected] (M. Mazzanti). http://dx.doi.org/10.1016/j.reseneeco.2015.04.003 0928-7655/© 2015 Elsevier B.V. All rights reserved.

186

A. D’Amato et al. / Resource and Energy Economics 41 (2015) 185–201

1. Introduction Many countries or regional areas are experiencing severe problems with respect to their waste management systems, due to a lack of disposal capacities and deficiencies in sorted waste collection and recovery (Mazzanti and Montini, 2014). This situation has led to detrimental effects on human health, worsened environmental quality, and a perceived reduction in the quality of public goods in urban settings. As clarified by theoretical and empirical investigations within the ‘economics of waste’ (Mazzanti and Montini, 2009; Pearce and Brisson, 1995; Shinkuma and Managi, 2011, among others), these problems seem to be driven primarily by policy failures and the lack of new and diversified investments in waste management and disposal facilities. Our paper intends to offer further progress in understanding these issues with reference to urban waste, focusing specifically on the case of Italy. Contrary to hazardous waste, whose illegal management involves trading (generally from the north to the south of Italy – legal flows usually move the other way around due to a lack of disposal capacity and insufficient recovery in the south), urban waste crises are driven by idiosyncratic regional features. Urban waste does not travel long distances in normal situations.1 A prototype of the “waste crisis” has primarily affected some southern regions of Italy in the last decade. For example, a well-known waste conflict has afflicted the metropolitan area of Naples, where more than 4 million people coexist with agriculture, food production, industry, waste treatment and disposal activities (D’Alisa et al., 2010). The related complexity, together with the strikingly different environmental and economic performances across areas in the country, has created problems regarding both the management of local ‘hot spots’ (Pasotti, 2010) and the positioning of waste disposal infrastructures (Jenkins et al., 2004). State and market failures (particularly a lack of adequate and enforced waste management and disposal programs, a lack of diversification, and an absence of environmental policies that reflect social and environmental values) were both present as pre-conditions of such an ongoing crisis. Beyond their still under-investigated social and sanitary consequences,2 such crises create a persistent emergency that favours criminal activities and mafia-type organizations.3 Indeed, as suggested by Legambiente (League for the Environment, the largest Italian NGO working in this field), the illegal waste business in Italy has increased tremendously over the years, reaching a turnover of approximately 7 billion euros in 2009 (Legambiente, 2010), and millions of tons of hazardous waste find their way outside legal management yearly. Criminal business affects legal and illegal markets: for example, organized crime is interested in managing illegal landfills (illegal disposal) and legal landfills (by buying under-priced land beforehand and distorting waste management towards landfilling), which generate higher rents than sorted waste collection.4 Illegal activities concern all types of materials, including urban garbage (Massari, 2004).

1 Although rising over time, the amount of international waste trade is a small part of total waste production (Mazzanti and Zoboli, 2013). For example, in 2010, the total generation of waste from economic activities and households in the EU27 amounted to 2570 million tons (of which 3.7% hazardous), while traded waste is around 11 million tons total, largely intra-EU. 2 Evidence of the worsening human health problems can be found, for example, in the several studies focusing on the correlation between increasing cancer rates and the presence of legal and illegal landfill sites in the Campania region (D’Alisa et al., 2010). 3 To simplify, we use the term “mafia” to indicate dominant, monopolistic organized crime groups located in the Italian territory, including, among others, the Camorra (Campania), the ‘Ndrangeta (Calabria), and the Sicilian Mafia itself. 4 The mafia may have an incentive to hamper the functioning of collection systems supporting institutional failures, creating social chaos around waste management. These actions are aimed at supporting landfilling as the only waste disposal and management option and, more generally, at favouring a lack of diversification in waste management and disposal. In such a way, organized crime can benefit from an enhanced territorial control capacity, as well as from increasing rents from land to be devoted to the creation or enlargement of landfills and to the storage of waste (D’Alisa et al., 2010). Despite association of the recent ‘waste crisis’ with Naples and the Campania region’s collapse of waste management, many areas in the centre and south of Italy suffer from lagging performances that have often worsened over time (Mazzanti and Montini, 2014). A new crisis might for example arise in the region where Rome is located (Lazio). Even there, though explicit criminal activities are not present – the boundary is often subtle – one single landfill monopolizes the market. Gate fees in Italy are among the highest in the EU (while landfill taxes are low, EEA, 2013, p. 25), as a consequence of the scarcity of usable land and of a lack of diversification, and attract criminal activities.

A. D’Amato et al. / Resource and Energy Economics 41 (2015) 185–201

187

Despite the potential relevance of various criminal activities in the waste management system, few theoretical and empirical studies have addressed the issue. Past evidence and attention were devoted to the core drivers of regional waste performance, such as income, population density and policy among others (Mazzanti et al., 2008). Nevertheless, other local idiosyncratic and spatial effects have been somewhat omitted from previous analyses. Our research questions concern how accounting for the presence of crime (mafia) in a local area might affect waste management choices and performances. Specific attention is devoted to sorted waste collection and landfilling, embedded in economic, policy and geographical settings. In the first part of the paper, we develop a simple model where, given the amount of produced waste, a representative consumer chooses whether to sort it according to the related cost. This cost depends on the choice of a waste management firm, which also decides how much of the unsorted waste is disposed of legally (determining illegal disposal as a residual). Finally, a waste authority sets the enforcement effort to reduce illegal disposal, and waste policy, in the form of a waste tax, is taken as exogenous. An added value of our analysis is the explicit comparison of the model above with one where organized crime is included; the mafia is assumed to extort a fee from the waste management firm based on the amount of waste disposed of illegally. Our theoretical analysis identifies two types of effects related to changes in (exogenous) waste policy: a direct one, which influences only legal unsorted waste disposal, and indirect ones, through the effect of the tax on the expected fine and (when the mafia is present) on the extortion fee in equilibrium. We also conclude that the presence of organized crime may worsen waste-related performance. In the second part of the paper, an empirical analysis is structured on a unique and very detailed balanced panel dataset covering 103 Italian provinces between 1999 and 2008. We take Italy as a relevant example due to the data availability and the pertinence of the issue, though our investigation is more general in its purpose. Specifically, we find that legal unsorted waste disposal is reduced and sorted collection is increased in provinces where waste policy is observed to switch from lump sum to incentive based,5 and organized crime worsens the performance in terms of waste management and disposal. The paper is organized as follows: Section 2 briefly reviews the relevant literature, Section 3 outlines the theoretical model and results, Section 4 introduces the empirical analysis, presenting the dataset and methodology we adopt, Section 5 shows the empirical results and provides comments on economic significance, and Section 6 concludes. 2. Relevant literature Under a theoretical point of view, we touch upon two strands of literature. First, we refer to papers related to waste policy in the presence of illegal disposal, particularly the studies by Sullivan (1987) and Fullerton and Kinnaman (1995). The latter conclude that the optimal policy involves a depositrefund system: a tax on the total output coupled with a rebate on proper disposal and recycling. Choe and Fraser (1999) then explicitly account for monitoring costs and identify the second-best optimal policy.6 Shinkuma (2003) also considers the effect of transaction costs on the choice of a suitable tool needed to achieve second best. We also refer to the literature on the economics of organized crime. Grossman (1995) models organized crime as a competitor of the State in the provision of public services. He shows that the existence of the mafia constrains a government’s behaviour. A similar trade-off is likely to arise in waste management choices. Another contribution worth mentioning here is Garoupa (2000), who accounts for organized crime in an optimal law enforcement model. The criminal organization is modelled as a vertical structure where the principal (the mafia) extorts rents from economic agents willing to act

5 Details on the changes from a waste tax to a waste tariff system in Italy will be given in presenting data. The “old” (before 1997) tax was calculated on the size of household living spaces, while the tariff is based on principles of full-cost pricing for waste management services and delivers some market-based incentives to the system, albeit its implementation has been slow and hampered by significant hurdles in the time period under scrutiny. 6 In a recent contribution, Ino (2011) considers illegal disposal both by households and by firms operating in the recycling market, showing how the second best optimal policy design is affected.

188

A. D’Amato et al. / Resource and Energy Economics 41 (2015) 185–201

illegally. In Garoupa’s paper, the existence of the mafia allows the government to save on enforcement resources because it makes illegal behaviour less attractive (i.e., it increases the related costs). In a subsequent paper, Chang et al. (2005) extend Garoupa’s framework by allowing for the coexistence of individual crime and organized crime; the authors remove the assumption that the mafia has a monopolistic power on criminal activities and make the “crime market structure” endogenous. The role of the mafia in the waste cycle is analysed in D’Amato and Zoli (2012), who conclude that under certain circumstances, a criminal organization operating in the waste cycle and extracting rents through socially harmful extortion might lead to higher levels of production and lower levels of enforcement effort. We borrow the modelling strategy from Garoupa (2000) and D’Amato and Zoli (2012), particularly by retaining the assumption that illegal waste disposal is “regulated” by organized crime. Turning to the empirical literature, several papers concentrate on waste generation and disposal drivers, focusing on the analysis of regional frameworks and/or on the role of policies (Hage and Soderholm, 2008; De Jaeger and Eyckmans, 2008; Dijkgraaf and Gradus, 2009, 2004; Kinnaman and Fullerton, 2000; Allers and Hoeben, 2010, among others) and, more in general, on the determinants of waste performance at the EU level (Mazzanti and Zoboli, 2009; EEA, 2007, 2009) and at the OECD level (Johnstone and Labonne, 2004). A paper that is somewhat complementary to ours is that of Almer and Goeschl (2013), who address illegal waste disposal in the German state of Baden-Wurttemberg and investigate how effective criminal law is in deterring waste-related offences.7 However, we are not aware of studies that bring together waste policy and organized crime issues in a joint theoretical–empirical way to better understand the determinants of waste performance. 3. Theoretical analysis 3.1. The model The aim of our theoretical analysis is to model waste-sorting decisions by a representative consumer (third and final stage), the choices concerning legal and illegal disposal by a waste management firm (second stage) and, finally, enforcement efforts by a waste regulator aiming to reduce the social costs related to illegal disposal and enforcement (first stage). When the mafia is present, it interferes with enforcement and (exogenous) waste policy by extorting a fee from the waste management firm (WMF). The amount of the fee is set by the mafia (when it is present) between the first and second stages. Starting from the last stage, a representative consumer produces y units of waste; a part of this waste is sorted (label the corresponding amount as s), depending on the related waste-sorting costs. We assume that s is a function of the effort chosen by the WMF, which we label as r, to reduce sorting costs for individuals. As is reasonable, we assume s (r) > 0 and further normalize s(r) = r so that the amount of sorted waste is exactly given by r, and the corresponding amount of unsorted waste is y − r. Moving backward, the WMF chooses r and the amount of unsorted waste that has to be disposed of legally (g) or illegally (b) to solve the following minimization problem: ming,r C(g, b, r) = (g) + ˇ(r) + tg + X(y − r − g)

(1)

Obviously, unsorted illegal disposal is derived as a residual (i.e., all unsorted waste that is not legally disposed of: b = y − r − g). The objective function in (1) is the sum of the costs related to the legal disposal of unsorted waste plus costs related to sorted waste, plus the total amount paid in the form of taxes on legal unsorted waste disposal, plus the expected payment for illegal disposal, which is given by the amount of illegal disposal multiplied by a unit expected payment, labelled as X. Notice that the WMF is modelled in a rather abstract way, particularly in terms of the separability between the costs related to legal disposal and those of sorted waste.

7 Almer and Goeschl (2013), among others, contribute to a stream of empirical literature focused on environmental enforcement (Gray and Shimshack, 2011 for a survey), specifically addressing waste. The bulk of contributions generally support that monitoring and enforcement are effective in increasing compliance with environmental regulations.

A. D’Amato et al. / Resource and Energy Economics 41 (2015) 185–201

189

The unit tax on legal disposal is, in our setting, a broad measure of the strictness of waste-related policies, taken as exogenous and labelled as t.8 This is coherent with our empirical strategy: one of our aims is to estimate how sorted and landfilled waste change in reaction to changes in waste policy strictness, and the theoretical model is expected to specifically capture induced policy effects (related to enforcement and to the mafia). The value of the expected payment for illegal disposal depends on whether the mafia is present in the waste cycle. More specifically,



X=

f

when the mafia is not present

f + x

when the mafia is present

(2)

where  is the probability that the illegal act is punished, which is a measure of the strictness of the monitoring effort on the WMF, f is the unit fine, and x is the fee extorted by the mafia when it is present. An implicit assumption in our model is related to the role played by the mafia. As outlined in the introduction, we rely on a monopolistic criminal organization, as do Garoupa (2000) and D’Amato and Zoli (2012), among others. In such a setting, potential offenders cannot act illegally on their own but rather have to pay a fee to the criminal organization. Though the monopolistic assumption is suggested as being a limiting one in some illegal sectors where mafia-type organizations typically operate (such as bootlegging, drug dealing and prostitution; see Chang et al., 2005), we deem it a realistic hypothesis in the waste sector, at least at a local level.9 Private management costs (e.g., transport costs, waste sorting) are given by functions (g) and ˇ(r) for legal disposal and sorted collection, respectively. Both functions are assumed to be strictly increasing and strictly convex in their arguments and satisfy (0) = 0 and  g (0) = 0 (respectively ˇ(0) = 0 and ˇr (0) = 0). Finally, private illegal disposal costs are assumed to be negligible and are normalized to 0. First-order conditions for the WMF are therefore as follows: g (g) + t = X, ˇr (r) = X

legal unsorted waste

sorted waste

(3) (4)

When the mafia is present, it chooses the extortion fee to maximize the rent from illegal waste disposal, accounting for the reaction of the WMF. The rent is given by the amount of illegal disposal multiplied by the fee (x)10 : maxx x(y − r − g)

(5)

In the first stage, a waste authority sets the enforcement effort to minimize the social costs related to illegal disposal and to monitoring and enforcement activities, anticipating the reaction of the mafia (when present) and of the WMF.11 The waste authority solves the following problem: min,f ı(b) + ()

(6)

subject to a standard constraint on the maximal fine that can be set up (i.e., f ≤ F, where F is the - exogenous - maximum feasible fine)12 .

8 In the case of Italy, a stronger waste policy commitment has taken, in the recent past, the form of locally managed more pervasive incentive based waste tariffs (Mazzanti et al., 2008). This will be discussed in presenting data. 9 Although the structure of the illegal disposal sector might be more complex, we keep our theoretical model tractable by limiting our attention to the role of the mafia as the “controller” of the illegal disposal facilities. 10 The criminal organization is assumed to be a rent maximizing entity which cannot be controlled by the government. The latter exerts enforcement effort towards the WMF only. See, in this respect, the discussion on the irrelevance of sanctions allocation in Garoupa (2000), based on Shavell (1997). 11 A leader–follower setting in the strategic waste authority-mafia interaction has been deemed as the most suitable to describe the specific waste policy context, where the criminal organizations react to the regulatory framework (D’Amato and Zoli, 2012). 12 The existence of a maximum possible fine can be justified on the basis of the existence of limited liability, equity reasons etc. (Polinsky and Shavell, 2000).

190

A. D’Amato et al. / Resource and Energy Economics 41 (2015) 185–201

Table 1 Comparative statics, no mafia. gt = − 1gg < 0 f

g =

gg

gf =

 gg

>0 >0

1 gg

>0

b = −f

gg +ˇrr gg ˇrr

<0

bf = −

gg +ˇrr gg ˇrr

<0

bt = −gt =

rt = 0 r =

f ˇrr

rf =

 ˇrr

>0 >0

The social costs of illegal disposal are given by a strictly increasing and convex function ı(b) and are to be interpreted in a broad way, for example, in terms of the additional damage related to disposing of waste illegally rather than legally.13 Finally, enforcement costs are given by the function (), which increases with the monitoring effort and satisfies (0) = 0,   (0) = 0 and   ≥ 0. As it is standard, we assume that the fine is a net social transfer and thus does not generate social costs.14 3.2. When organized crime is not present Setting X = f, and from the first-order conditions (3) and (4), we obtain the comparative statics summed up in Table 1.15 There is no direct effect of unsorted waste policy on sorted waste. This is indeed a consequence of the assumptions made concerning the shape of (1)—primarily cost separability. However, we will show below that indirect effects arise through the link between the exogenous unit tax and the monitoring effort in equilibrium. Anticipating the effect of its monitoring and enforcement decisions on the WMF’s choices, the waste authority chooses  and f to minimize (6). Because the unit fine is a net social transfer, and because (6) can easily be shown to be strictly decreasing in f, we reach the standard conclusion that the fine is set at the maximum possible level (i.e., f = F) to save resources (Polinsky and Shavell, 2000). The corresponding first-order (necessary and sufficient) conditions for the monitoring effort in the absence of the mafia are: ıb (·)b +  (·) = 0

(7)

Superscript n labels equilibrium variables in the no mafia circumstance where needed. Straightforward comparative statics imply in this case: dn ıbb b bt =− > 0. dt ıbb b2 + 

(8)

As expected, the monitoring effort increases with the unit tax rate. 3.3. When organized crime is present When the mafia is present, the WMF once more solves problem (1) with respect to g and r, but X = f + x, where x is the unit extortion fee extracted by the mafia. Comparative statics closely follow those obtained in the absence of the mafia; however, we can witness how sorted waste and legal unsorted waste disposal change with the unit extortion fee (see the first row of Table 2).16 As 13 The benefits from sorted waste are not explicitly included in (6). This is not expected to alter results in the very simple framework set out in this paper. A more general treatment of welfare is out of the scope of this theoretical analysis and will be the subject of further research. 14 We will limit, throughout the theoretical model, to interior solutions, both in the absence and in the presence of the mafia (with the exception of the unit fine - see below). Additional assumptions are made to obtain readable insights and to make matters as simple as possible; namely, in what follows we will normalize y = 1 and assume that the second derivatives of all convex functions are constant. Also, second order conditions are straightforward and not shown for the sake of brevity. 15 We will label the first (second) derivative of z with respect to x as zx (zxx ). All details concerning comparative statics reported in Tables 1 and 2 are in Appendix B1, available online. 16 We also recall our assumption of constant second derivatives (see footnote 14). Again, Appendix B1, available online, provides details on comparative statics and on calculations.

A. D’Amato et al. / Resource and Energy Economics 41 (2015) 185–201

191

Table 2 Additional comparative statics in the presence of the mafia. gx =

1 gg

x =

− 2f

>0 <0

rx =

1 ˇrr

>0

xf = − 2 < 0

gg +ˇrr gg ˇrr

<0

ˇrr 2(ˇrr +gg )

>0

bx = − xt =

expected, ceteris paribus legal behaviour increases and illegal behaviour is reduced by an increase in extortion. Anticipating the effect of x on the choices of the WMF, the mafia chooses the extortion fee to solve problem (5). The related first-order conditions lead to the comparative statics results reported in the second row of Table 2, showing that the extortion fee decreases with monitoring and increases with taxation. This is reasonable: an increase in monitoring makes illegal disposal more costly (in expected terms), forcing the mafia to reduce the related “price”. On the contrary, an increase in the tax on legal unsorted waste disposal increases the WMF’s willingness to pay for illegal disposal. The waste authority again minimizes (6) subject to WMF and mafia first-order conditions. The unit fine is, also in this case, a net social transfer, and (6) can be shown to be strictly decreasing in f. The fine is therefore set, again, at the maximum possible level (i.e., f = F). First-order (necessary and sufficient) conditions for the waste authority are: ıb (·)(b + bx x ) +  (·) = 0

(9)

Superscript m labels equilibrium variables under the mafia where needed. Given the properties of the ı(·) and (·) functions, it is easily shown that it is not possible for (7) and (9) to hold at the same time if monitoring effort levels satisfy m ≥ n .17 In other words, to get interior solutions both in the absence and in the presence of mafia we need m < n . This is linked to monitoring being less effective in reducing illegal disposal under the mafia, which is in turn due to the mafia reacting to increases in  with partially counterbalancing decreases in x. Additionally, from comparative statics, we can conclude that18 : dm ı (b + bx x )(bt + bx xt ) ıbb b bt = − bb =− > 0. 2 2 dt ıbb (b + bx x ) +  ıbb b + 4

(10)

The result obtained in the absence of the mafia is therefore confirmed: an increase in the tax rate implies a stricter monitoring effort. On the other hand, comparing (8) and (10), it is straightforward to show that dm /dt < dn /dt: the presence of organized crime implies a lower reactivity of the monitoring effort to changes in the tax rate on legal unsorted waste disposal. Therefore, the presence of organized crime affects g, r, and b and their reactivity to waste policy through a weaker link between  and t and through the extortion fee; these considerations suggest a potentially relevant effect of the mafia’s presence on waste-related performance. 3.4. Model results From comparative statics results, we can first of all assess the net effect of changes in the stringency of waste management policy, which is measured in our setting by the tax rate on legal unsorted waste disposal. In the absence of the mafia, from the comparative statics results in Table 1 and from (8), we have r (dn /dt) > 0, whereas in the presence of the mafia, the results in Table 2 and (10) clearly

17 Recall, from footnote 14, that we are limiting our attention to interior solutions. A detailed proof of this conclusion is available online (Appendix B2). 18 Using results in Tables 1 and 2, the second equality in (10) stems from: b + bx x = b /2 and bt + bx xt = bt /2.

192

A. D’Amato et al. / Resource and Energy Economics 41 (2015) 185–201

show that the overall effect of a change in t is: (r + rx x )(dm /dt) + rx xt , which is, again, positive, as r + rx x = (F/2ˇrr ) > 0. Testable implication 1. The amount of sorted waste increases with the stringency of waste management policy. Clearly, the purposely stylized features of our model do not allow us to cover all the potential impacts of waste policies on sorted collection. Rather, the (indirect) impact of policies through the expected cost of illegal disposal can be seen as a subset of the broader linkages which are suggested by the existing literature (e.g. Kinnaman and Fullerton, 2000; Isely and Lowen, 2007). Turning to the net effect of changes in t on the level of legal unsorted waste disposal, we have to account for the negative direct effect through which an increase in t decreases g, ceteris paribus. This direct effect, which triggers a corresponding increase in illegal disposal, has to be compared with the indirect effects driven by the monitoring effort and by the extortion fee when the mafia is present. It can be demonstrated that in the absence of the mafia, gt + g (dn /dt) < 0, and in the presence of the mafia, gt + gx xt + (g + gx x )(dm /dt) < 0.19 In other words, the net effect is always negative: the direct effect of a stricter waste policy on legal unsorted disposal always dominates the indirect effects in equilibrium. As a result: Testable implication 2. The amount of legal unsorted waste disposal decreases with the stringency of waste management policy. Again, Testable implication 2 should be interpreted in light of our model: a stricter policy implies weaker (stronger) incentives towards legal unsorted (illegal) disposal, which are only partially offset by the increase in enforcement effort and in the extortion rate (when the mafia is present). We can therefore conclude that the reduction of legal unsorted disposal is, at least in a significant part, not good news, due to illegal disposal.20 Finally, it must be noted again that, as in the case of the policy impact on sorted waste, our theoretical analysis only focuses on a specific link between policy and unsorted waste disposal, namely the one related to illegal disposal incentives, enforcement and extortion. We can now turn to the effect of the presence of a criminal organization on waste-related choices by the WMF. From first-order conditions (3) and (4), it is clear that given t, the levels of sorted waste and legal unsorted disposal are uniquely determined by the unit expected payment for illegal disposal, which in turn depends on the chosen monitoring effort and on the extortion fee imposed by the mafia (when it is present). As shown in Section 3.3, m < n . On the other hand, the net effect also depends on the increase in the payment by the WMF due to mafia extortion. A smaller payment under the mafia would result if n F > m F + x. If mafia implies a reduction in the expected fine which is larger, in absolute terms, than the extortion fee, then the incentives towards legal forms of waste disposal and management are reduced in the presence of organized crime. It can be shown, by putting additional structure on our model, that a smaller unit payment for illegal disposal under the mafia is indeed possible.21 This takes us to our last testable implication. Testable implication 3: The amount of sorted waste and legal unsorted waste disposal can be lower in the presence of the mafia than when organized crime does not enter into the waste cycle.

n

2

19 More specifically, it can be shown that gt + g d = −( − ıbb r b )/(gg (ıbb b +  )) < 0 and dt gt + gx xt + (g + gx x )(dm /dt) = − (2(ˇrr + 2 gg )  +  gg ıbb b 2 )/( gg (ıbb b 2 + 4  )( gg + ˇrr )) < 0. 20 Indeed, equilibrium illegal disposal is shown to increase with t, as: bt + b (dn /dt) =   /( gg (ıbb b 2 +   )) > 0 and bt + bx xt + m 2 (b + bx x ) ddt = 2 /(gg (ıbb b + 4 )) > 0. 21 This conclusion can be derived by assuming specific quadratic functional forms for the (·), ˇ(·), (·) and ı(·) functions, and restricting parameter values in such a way to guarantee strictly positive values for all equilibrium variables and 0 <  < 1 under any scenario. The complete proof is available online (Appendix B3).

A. D’Amato et al. / Resource and Energy Economics 41 (2015) 185–201

193

4. Empirical framework 4.1. The data and the model The analysis uses the yearly editions of the Italian environmental agency’s waste reports (available online at http://www.isprambiente.gov.it/it/pubblicazioni/rapporti). These reports provide a rich set of waste management and disposal data, including data on municipal solid waste (MSW) that has been sorted and landfilled in all the Italian provinces,22 during the 1999–2008 period. We merge these data with official data on provincial level socio-economic drivers (source: Italian National Statistical Agency-ISTAT), such as value added as a proxy of provincial income, population density and tourist-related flows, which become a crucial factor considering that they add opportunity costs to the effects of density. Waste performance differs widely among Italian provinces, making the provincial level of analysis the most suitable one. Though northern Italy is rapidly evolving towards a high level of sorted waste collection, which peaks at approximately 75% in some provinces, the average figure for the country is still dominated by landfilling (EEA, 2013). Further, even some northern regions suffer from landfill criticality. The specification tested in the panel-based analysis is as follows (Johnstone and Labonne, 2004; Mazzanti and Zoboli, 2009): Log(waste)it = ˛i + tit + ˇ1 log (economic driver)it + ˇ2 log (socio − economic factors)it + ˇ3 (environmental policy)it + ˇ4 (organized crime variable) + εit

(11)

where ˛i are provincial fixed effects, tit is a time trend, and εit is the error term. Waste indicators that are specified as dependent variables are the collection of sorted waste and landfilled waste. 4.2. Waste, economic, and tariff indicators In (11), waste indicators—here, sorted waste collection and landfilling of MSW—are introduced in per capita terms. The term ˇ1 refers to the main economic driver (value added per capita at the provincial level; VA).23 Other socio-economic factors are added to the core specification. In our model, these include population density (DENS) and tourist numbers (TOURIST). Population density may control for different land values. We therefore expect this factor to be positively related to landfill diversion. The third term (ˇ3 ) refers to waste management/policy: we take the share of provincial municipalities and the provincial population covered by a new ‘waste tariff’ regime, which substituted the old ‘tax’ regime. This variable varies across time and space. It is in essence a cost factor associated with waste production and disposal, with some internal elements that may reduce the imposed price if waste reduction (e.g., composting) is implemented. The ‘new’ waste management tariff was introduced by Italian Law No. 22/1997 and in theory should have fully substituted the former waste management tax. A mechanism not much different from the older tax, however, has been kept in force in many Italian municipalities. Law 22/1997 provided for a transitional phase that has proven to be quite gradual and slow. The tax was calculated on the size of household living spaces, whereas the tariff was based on principles of full-cost pricing for waste management services and delivered market-based incentives to the system (see Mazzanti et al., 2008). The effective implementation of the tariff system remained highly dependent on local policy decisions and practices. The observation of an early implementation of the new tariff-based system, therefore, may be a sign of a stronger waste policy commitment. We note that implementation is heterogeneous even across areas with similar per capita incomes. We use two variables: tariff diffusion in terms of the provincial population share that is covered (TARPOP) and the share of municipalities that have implemented the new system (TARMUN), the former capturing the role of tariff implementation in large metropolitan areas (for example, Rome). 22 23

Provinces are the intermediate administrative level between regions (20) and municipalities (around 8000). We also test a squared value added term, in order to account for non-linearity, which is not significant.

194

A. D’Amato et al. / Resource and Energy Economics 41 (2015) 185–201

It must be noted that several factors may give rise to endogeneity concerns. Simultaneity, for instance, may be something to consider due to the nature of the waste policy effort. On the one hand, regulations may be an important explanatory variable of waste management performance, but on the other, it is reasonable to assume that provinces with a relatively worse waste management performance may have implemented more stringent policy measures in the analysed period compared to the more virtuous ones to fill the gap. Finally, our variables TARPOP and TARMUN only partially account for the actual stringency of waste management policies. This phenomenon may generate a problem generally known as measurement error, which, according to econometric theory (Wooldridge, 2001), causes a bias in OLS estimates. From an econometric perspective, these aspects will be addressed through an instrumental variable (IV) estimator. 4.3. Organized crime indicators Variables related to organized crime have been created primarily thanks to an existing database of the Ministry of Internal Affairs, which shows all the municipal governments that have been dissolved in the last two decades due to guilty verdicts concerning mafia connections. Government officers manage the city for a few years (usually 2–3), and then, new elections are called. We have aggregated information at the provincial level in coherence with provincial waste data.24 One other hypothetical concern is related to the possible omission of relevant variables that might be correlated to our dependent variables and generate a potential bias in regression estimates. To clarify this point, we refer again to our theoretical model: indeed, we suggest enforcement and/or extortion as possible channels through which policy strictness and the presence of the mafia can influence legal disposal and sorted waste. However, clearly, enforcement levels and waste crime can depend on factors other than those just outlined, possibly implying distorted estimates. It is reasonable, however, to assume that given the short time span considered here, these elements can be accounted for as time invariant, and consequently, this bias is mitigated by the fixed effect (FE) estimation procedure. This reasoning is based on the plausible assumption that the determinants of enforcement strategies and illegal behaviour (other than mafia and policy), depend on “structural” features and are fixed through time (or at least slow to change). Additionally, we do not explicitly address incineration, which remains a small part of waste management options in Italy (EEA, 2013). In the analysis, different ‘mafia variables’ are tested. We include various specifications that feature either binary or continuous variables. A. binary variables. 1. A ‘narrow’ variable (CRIMEnarr), which, for provinces that include municipalities turned over due to mafia, takes value 1; 2. a ‘broad’ variable (CRIMEspill), in which neighbouring provinces are also considered, accounting for proximity and geographical spillover effects. We aim to capture negative spillover effects that may be of high importance in some waste production chains. In many cases, waste management bears witness to the cooperation of different local public agencies. We note that to account for the lagged and dynamic effect of organized crime on institutional and economic settings and performances, the variable assumes a value of one for the three years preceding the “institutionalization” of the mafia connection by the Ministry of Internal Affairs. We can reasonably hypothesize that mafia connections had been in place for a period of time before being recognized and sanctioned. After a local government is dissolved (when a conviction occurs), the value is 0. This does not mean that crime disappears, but rather that the local government has changed officials.

24 We stress that we aim to capture specific mafia effects on the functioning of local institutions, not general crime-related social factors.

A. D’Amato et al. / Resource and Energy Economics 41 (2015) 185–201

195

Table 3 Descriptive statistics and acronyms. Acronym

Variable description

Mean

Min

Max

SEPARATED COLLECTION LAND-WASTE

Separated collection (kg per capita) Municipal solid waste yearly Landfilled (kg per capita) Municipal solid waste collected (kg per capita) Provincial yearly value added per capita (base: Euro 2000) Population/surface (inhabitants/km2 ) Annual tourist attendances (per capita) Share of population living in municipalities that introduced a waste tariff substituting the former waste tax (%) Share of municipalities that introduced a waste tariff substituting the former waste tax (%) Presence of at least one municipality guilty of mafia connection inside the province Presence of at least one municipality guilty of mafia connection inside the province, or in a nearby province Share of provincial population living in municipalities guilty of mafia connection Electoral turnover share (at provincial Level, %) Employment/inhabitants Share of people aged 14 or more which participated in the last 12 months in ecological or peace associations

115.56 310.64

0.09 0

378.34 1,145.01

524.76 18,267.36

251.91 9,386.46

882.87 30,889.24

246.85 7.22 13.50

31.16 0.39 0

2,646.92 58.83 100

7.81

0

100

0.07

0

1

0.19

0

1

0.01

0

0.21

82.51 0.40 1.85

57.00 0.25 0.50

90.56 0.56 4.70

MSW VA DENS TOURIST TARPOP

TARMUN CRIMEnarr CRIMEspil

CRIMEsh SOCCAP EMPLOY ASSOC

B. continuous variable. This variable is represented by the population share of a province related to municipalities infiltrated by the mafia. This captures ‘intensity’ and is calculated as 3 years moving average (CRIMEsh). All variables are summarized in Table 3. 5. Econometric evidence We summarize here the main outcomes regarding sorted waste collection and landfill diversion (legally non-recyclable waste disposal). We refer to Tables 4–7. The outcomes relate to FE and IV regressions. For the IV estimations, we successfully exploit two ‘social capital’ indicators: the share of electoral turnout (Guiso et al., 2004) and membership in ecological and peace associations (Putnam, 1993). These are correlated with actions of local commitment to public good provision (e.g., policy actions) but not directly linked to waste performance. Regarding electoral turnout, provincial heterogeneity is striking in Italy. We consider membership in associations as a source of generalized trust and social ties conducive to governmental efficiency. Another instrument considered in the IV analysis is employment share. This variable may be a reasonable proxy of social cohesion (Marselli and Vannini, 1997, 2000). Finally, Tables 5 and 7 control for the time dimension of the panel through the inclusion of area specific time trends (Columns 1 and 2) and time dummies (Columns 3 and 4). In addition, a dynamic GMM specification controls for the persistency of the dependent variable and again tests for the inclusion of time dummies. Detailed sensitivity analyses on the construction of both policy and crime variables are presented in an online Appendix (Tables A1–A3). 5.1. Sorted waste 5.1.1. Main evidence Table 4 presents the results for the main specification (11). First, the value added generated in a province is positively related to waste performance. More notably, a denser population does not enhance sorted waste collection rates in Italy, a result in contrast to recent studies conducted on UK

196

A. D’Amato et al. / Resource and Energy Economics 41 (2015) 185–201

Table 4 Separated collection (main table). (1) FE

(2) FE

(3) FE

(4) FE

(5) 2SLS-IV

(6) 2SLS-IV

(7) 2SLS-IV

5.6655*** −0.7045 0.8786*** 0.0038** −0.0556

5.5289*** −0.2281 0.8574*** 0.0030

5.5493*** −0.2721 0.8669*** 0.0034*

5.4093*** −0.7170 0.8518***

4.0222*** −5.3303 0.6840* 0.0419**

4.0123*** −4.1574 0.7510** 0.0345***

4.3736*** −4.2489 0.8202*** 0.0342***

−2.1733*

−2.3031**

−2.6820**

0.0039***

0.0037***

0.0030**

Excluded Instruments

ASSOC

Sargan (p-value) Davidson Mackinnon Provincial FE Time trend N

0.000 0.000 Yes Yes 914

ASSOC, SOCCAP 0.1272 0.000 Yes Yes 914

ASSOC, SOCCAP, EMPLOY 0.1322 0.000 Yes Yes 914

Model VA DENS TOURIST TARMUN CRIMEnarr CRIMEspill CRIMEsh TARPOP MSW

−0.3815***

−0.3896*** −3.4821***

0.0044***

Yes Yes 1029

0.0044***

Yes Yes 1029

0.0043***

Yes Yes 1029

0.0036** 0.0043***

Yes Yes 1029

*

p < .1. p < .05. *** p < .01. Standard errors are cluster-robust by province. **

Table 5 Separated collection (robustness checks).

VA DENS TOURIST TARMUN CRIMEspill CRIMEsh MSW Lagged DEP VAR Excluded Instruments Sargan (p-value) Davidson Mackinnon AR(2) in first differences Provincial FE Time trend × area dummies Year FE N

(1) FE

(2) 2SLS-IV

(3) FE

(4) 2SLS-IV

(5) Difference GMM

4.1892*** −2.0654 0.4367* 0.0029** −0.2532***

3.9553*** −2.5731 0.4441 0.0246**

0.5270 −6.0016*** 0.1865 −0.0046 −0.2003***

0.0673 −5.1279*** 0.2103 −0.0182

−0.3746 1.4813 −0.0694 0.0171*

−1.6696** 0.0000

−2.8377* 0.0115*** 0.7277***

0.0019*

−1.0101 0.0017

0.0003

ASSOC, SOCCAP 0.3734 0.2439 Yes Yes No 1030

Yes Yes No 914

ASSOC, SOCCAP 0.5604 0.0099 Yes No Yes 1030

Yes No Yes 914

0.349 0.451 Yes No Yes 927

*

p < .1. p < .05. p < .01. Standard errors are cluster-robust by province. **

***

data (Abbott et al., 2011). On the contrary, tourist flows (TOURIST) are an important and statistically significant covariate in Italy that captures scale effects of waste management and opportunity costs of disposal without recovery. We also note that both waste policy variables, which capture the diffusion of tariffs, positively affect sorted waste collection. This conclusion refers to testable implication 1. This tariff is a strong sign of a local policy commitment.25

25 R2 and F statistics (of the provincial dummy) are not presented in the tables to preserve space. The F test always confirms the significance of the individual dummy, while R2 generally suggests a good model fit.

A. D’Amato et al. / Resource and Energy Economics 41 (2015) 185–201

197

Table 6 Legal disposal of waste (landfilling – main table). (1) FE

(2) FE

(3) FE

(4) FE

(5) 2SLS-IV

(6) 2SLS-IV

(7) 2SLS-IV

−0.9557 −3.1322** −0.8137** −0.0048** −0.5725**

−1.1876* −3.1392** −0.7415** −0.0045**

−1.1487* −3.2462** −0.7328** −0.0043*

−1.0035* −3.2064** −0.8264** −0.5544**

−0.9279 −3.5075** −0.8981*** −0.0056 −0.6472***

−1.3196** −4.6057*** −0.9401*** 0.0021 −0.6007***

−1.0123* −3.2911** −0.8564*** −0.0074* −0.5358***

−0.0023* −0.0005

−0.0006

−0.0006

−0.0004

Excluded Instruments

ASSOC

Sargan (p-value) Davidson Mackinnon Provincial FE Time trend N

0.000 0.9748 Yes Yes 818

ASSOC, SOCCAP 0.1566 0.9541 Yes Yes 818

ASSOC, SOCCAP, EMPLOY 0.1418 0.8012 Yes Yes 818

Model VA DENS TOURIST TARMUN CRIMEnarr CRIMEspill CRIMEsh TARPOP MSW

−0.0628 −0.2055* −0.0006

Yes Yes 926

−0.0003

Yes Yes 926

−0.0003

Yes Yes 926

Yes Yes 926

*

p < .1. p < .05. *** p < .01. Standard errors are cluster-robust by province. **

Table 7 Legal disposal of waste (landfilling – robustness checks).

VA DENS TOURIST TARMUN CRIMEnarr MSW Lagged DEP VAR Excluded Instruments Sargan (p-value) Davidson Mackinnon AR(2) in first differences Provincial FE Time trend × area dummies Year FE N

(1)

(2)

(3)

(4)

(5)

FE

2SLS-IV

FE

2SLS-IV

Difference GMM

−0.2291 −2.0330 −0.5985* −0.0031* −0.5613** −0.0005

−0.5259 −2.6967*** −0.7469*** 0.0064 −0.5902*** −0.0009

0.1322 −2.4724 −0.6845* −0.0034 −0.5548** 0.0001

0.1412 −3.7452*** −0.7108*** 0.0057 −0.5769*** 0.0004

3.6682 3.2659 −1.9255 −0.1846 −16.7217*** 0.0572 −0.2270

ASSOC, SOCCAP 0.8421 0.3502 Yes Yes No 927

Yes Yes No 818

ASSOC, SOCCAP 0.6630 0.2899 Yes No Yes 927

Yes No Yes 818

0.714 0.364 Yes No Yes 835

*

p < .1. p < .05. *** p < .01. Standard errors are cluster-robust by province. **

Regarding the effect of ‘organized crime-related variables’,26 we can state that where the mafia is present and/or more pervasive, sorted waste collection performance tends to be lower. This result confirms the possibility that the presence of organized crime worsens waste management performances (testable implication 3). Indeed, although on the one hand, the ‘narrower’ crime-related effect captured by CRIMEnarr27 is not significant in explaining sorted waste collection performance, the ‘wider’ 26

The correlation between crime covariates and tariff diffusion is negative and quite significant, but under 0.25 value. We recall that such crime dummies are time variant and assume a lag between the ‘event’ (organized crime is recognized) and the cause (mafia presence). We assume that the mafia exerts its effects 3 years before its presence is formally revealed by the State through the judiciary system. 27

198

A. D’Amato et al. / Resource and Energy Economics 41 (2015) 185–201

mafia factor (geographical spillover) in contrast is highly significant across specifications from both a statistical and an economic point of view (Columns 2 and 4).28 This is very reasonable considering that sorted waste and collection activities for separated collection and recovery are often characterized by strong links in the management procedures between local authorities and waste utilities in the form of associations and consortia of a public or mixed public–private nature. Currently, utilities that address waste management often operate beyond the scale of provinces or municipal consortia. Overall, policy and market developments have stretched the functioning of waste management markets beyond provinces. Thus, the spatial spillovers that characterize crime and specific mafia networks negatively affect the performance of waste management, which itself depends on the good management of the waste cycle: crime networks spill over across administrative and geographical jurisdictions.29

5.1.2. Instrumental variable regressions In Columns 5–7 of Table 4, we adopt an IV approach to our benchmark specification of Column 3, instrumenting the policy variable with: (i) the share of people who participated in ecological or peace associations (ASSOC); (ii) the provincial share of electoral participation (SOCCAP); and (iii) the employment rate (EMPLOY) (source: Italian National Statistical Agency-ISTAT). Following the literature on social capital and regulations (Ng and Wang, 1993; Hettige et al., 1996) and social capital and development, with a historical emphasis on Italy as a case study (Guiso et al., 2006; Putnam, 2001, 1993 [Chapter 6]; Tabellini, 2010), we strongly believe that social capital-cultural indicators may be a valid instrument (expected to be correlated with the policy effort and exogenous to the main relationship). The regression results for the instrumental variable estimations generally confirm previous results, and the F-test for the first stage is well above the usual cut-off level of 10. A large downward bias is observed in the OLS estimates, and the chosen instruments all have the expected sign (see first stage regressions in online Appendix A, Table A4) and appear to be truly exogenous, as shown by the Sargan tests presented in Table 4.

5.1.3. Robustness checks In order to control for the time dimension of our panel, in Table 5 we test the robustness of our results to the inclusion of area-specific time trends (Columns 1 and 2) and year dummies (Columns 3 and 4). The inclusion of area-specific trends does not alter the main evidence, even when controlling for more specific time effects. The coefficient of the crime variables turns out to be insignificant for the IV estimations of Column 2. With the introduction of time dummies, on the contrary, the effect of waste policies vanishes, whereas results for the mafia variables remain coherent with the theoretical framework. This is a reasonable result considering that waste management policies are captured by TARMUN, which is an indicator that is expected to increase through time. Finally, we implement fully dynamic models to provide further robustness checks for the core set of outputs (see Column 5, Table 5). Following Holtz-Eakin et al. (1988), and Arellano and Bond (1991), we use all valid lags of the untransformed variables (which means lags 2 and up) as instruments for the endogenous variables and all contemporaneous and lagged levels for predetermined regressors. The

28 It is worth noting that in all cases we carried out an additional sensitivity test by using standard crime variables (source: Italian National Statistical Agency-ISTAT), such as ‘criminal offences on a per capita provincial basis’. We highlight the very similar economic and statistical significance of this variable, which in any case restricts the time span to 2000–2005 due to limited data availability. Results are available upon request. 29 As an additional test, we separate the CRIMEspill variable into its two components: CRIMEnarr and a dummy that takes on a value equal to one only if the mafia is present in adjacent provinces. Results (available upon request) show that only the adjacent component is statistically significant (t = −4.16) and associated with a negative coefficient (−0.47), confirming our idea that influence from adjacent provinces is relevant.

A. D’Amato et al. / Resource and Energy Economics 41 (2015) 185–201

199

outcomes primarily confirm the results obtained.30 We note that the lag is significant and that, among the covariates, TARMUN and organized crime do not change signs, compared to the FE/IV analyses.31 5.2. Legal disposal of waste 5.2.1. Main evidence Table 6 shows the results for ‘waste landfilled per capita’, i.e., legal disposal. Again, time trends are included in the main specifications. Economic and health-related opportunity costs largely explain the evidence for population density: a 1% increase in population density leads to a 3% increase in landfill diversion. Turning to the policy and crime variables, the effect of tariffs is negative, and the forces related to organized crime send robust messages. In particular, legal unsorted waste disposal decreases as a result of a more widespread ‘incentive based’ tariff. We cannot however exclude (as discussed in the theoretical model) that at least part of the reduction in legal unsorted disposal translates into more illegal disposal. Regarding CRIMEnarr, the likelihood that the structural presence of mafia networks increases landfill diversion is strong. The significance of the ‘narrow’ spillover effect is reasonable here given that landfill management, unlike recovery and sorted waste activities, is more circumscribed in defined territories, often within a municipal area. This is not a virtuous outcome at all: in the absence of official data on illegal disposal, we can indirectly infer that criminal activities specifically located in the province itself or in neighbouring ones reduce both sorted waste and legal forms of landfilling. In the face of an increasing trend in waste generation and in the absence of incinerators32 in the areas most affected by the mafia, we end up with a positive relationship between the presence of organized crime on the one hand and higher illegal disposal on the other. 5.2.2. IV estimations Also in the case of unsorted legal disposal, the policy variable can be simultaneous to the dependent variable. It is plausible to conclude that policy effort is more stringent where waste management is more distant to targets. For this reason, in Columns 5–7 of Table 6, we instrument the TARMUN variable. In this case, we use column 1 as benchmark specification. IV results are in line with fixed effect estimations. It must be noted, however, that the Davidson-Mackinnon test does not reject the null hypothesis, which indicates that endogenous regressors’ effects on the estimates are not meaningful, and an OLS estimator of the same equation would provide consistent estimates. 5.2.3. Robustness checks Table 7 tests the robustness of results to the inclusion of area-specific time trends (Columns 1 and 2) and year dummies (Columns 3 and 4). It is worth noting that crime-related effects are not absorbed by time components, and they remain statistically significant in the more demanding specifications of Table 7. Concerning the GMM-diff estimator, in this case we also use all valid lags of the untransformed variables as instruments for the endogenous variables, starting from the third lag, and all contemporaneous and lagged levels for predetermined regressors.33 The lagged dependent variable is not statistically significant in this case, suggesting that landfill diversion did not generate persistency.

30 We report two tests: the Arellano–Bond test of autocorrelation which in both cases does not reject the null hypothesis of absence of autocorrelation, and the Sargan test for over-identifying restrictions, which confirms that the instruments appear exogenous. 31 Differences in differences (DID) estimation mechanisms confirm that after the introduction of the tariff, separated collection significantly increases and landfilling of waste decreases. If we add the usual set of regressors to the baseline DID model, the outcome is unaffected. Results are available upon request. 32 For more insights on the Italian waste management sector see Mazzanti et al. (2011). 33 According to the result of specification tests.

200

A. D’Amato et al. / Resource and Energy Economics 41 (2015) 185–201

6. Conclusions We analyse how waste management and disposal performances in highly decentralized institutional settings are influenced by organized crime activities and waste management policy, two locally developed and idiosyncratic factors. We first develop a theoretical model in which the role of the mafia in the waste cycle is explicitly addressed and where indirect effects of waste policy through changes in the enforcement effort are included. We then outline that the presence of organized crime may worsen waste-related performance. Econometric analyses that exploit provincial panel data show that landfill diversion and sorted waste are positively affected by the local diffusion of waste tariffs that aim at achieving full cost recovery and introducing economic incentives, even if we cannot exclude (at least part of) the reduction of legal disposal to generate increases in illegal disposal. The presence of the mafia in local governments hinders legal forms of waste management and disposal. Another interesting insight is the existence of significant spatial phenomena related to organized crime: in the case of sorted waste, organized crime seems to exert effects from outside the immediate administrative area of relevance. The landfill of waste and crime are instead characterized by more locally circumscribed effects. Overall, the paper shows the relevance of very decentralized territorial aspects for waste management and disposal. We here investigate the key policy effort and crime factors. The main novelty of the paper is the role of organized crime within local municipalities, which emerges as a strong obstacle to achieving better waste performance. Further research could intensify the investigation of spatial spillovers related to both crime and policy issues. Acknowledgment The authors would like to thank the editor, two anonymous reviewers, Annalisa Fabretti, Mariangela Zoli and seminar audience at the University of Kiel for precious comments and suggestions. Previous versions of this paper have been presented at the 2011 ENVECON Conference, London, UK, March 2011, at the 2011 ESEE Conference, Istanbul, Turkey, June 2011, at the EAERE 2011 Pre-Conference on Waste Economics, Rome, Italy, 29 June 2011, at the XXIII Annual SIEP Conference, Pavia, Italy, September 2011, at the 52nd Annual SIE Conference, Rome, Italy, October 2011 and at the Workshop on The Economics of Waste Management - Bath, UK, 10 July 2012. The usual disclaimers apply. Appendices A and B. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.reseneeco.2015.04.003. References Abbott, A., Nandeibam, S., O’Shea, L., 2011. Explaining the variation in household recycling rates across the UK. Ecol. Econ. 70, 2214–2223. Almer, C., Goeschl, T., 2013. The Sopranos redux: the empirical economics of waste crime. Reg. Stud. (Epub ahead of print). Allers, M., Hoeben, C., 2010. Effects of unit based garbage pricing: a differences in differences approach. Environ. Resour. Econ. 45, 405–428. Arellano, M., Bond, S., 1991. Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations. Rev. Econ. Stud. 58, 277–297. Chang, J.J., Lu, H.C., Chen, M., 2005. Organized crime or individual crime? Endogeneous size of a criminal organization and the optimal law enforcement. Econ. Inq. 43, 661–675. Choe, C., Fraser, I., 1999. An economic analysis of household waste management. J. Environ. Econ. Manage. 38, 234–246. D’Alisa, G., Burgalassi, D., Healy, H., Walter, M., 2010. Conflict in Campania: waste emergency or crisis of democracy. Ecol. Econ. 70, 239–249. D’Amato, A., Zoli, M., 2012. Illegal waste disposal in the time of the mafia: a tale of enforcement and social well being. J. Environ. Plan. Manage. 55, 637–655.

A. D’Amato et al. / Resource and Energy Economics 41 (2015) 185–201

201

De Jaeger, S., Eyckmans, J., 2008. Assessing the effectiveness of voluntary solid waste reductions policies: methodology and a Flemish case study. Waste Manage. 28, 1449–1460. Dijkgraaf, E., Gradus, P., 2004. Cost savings in unit based pricing of household waste: the case of The Netherlands. Resour. Energy Econ. 26, 353–371. Dijkgraaf, E., Gradus, P., 2009. Environmental activism and the dynamics of unit based pricing systems. Resour. Energy Econ. 31, 13–23. EEA, 2007. The Road from Landfilling to Recycling: Common Destination, Different Routes. European Environment Agency, Copenhagen. EEA, 2009. Diverting Waste from Landfills. European Environment Agency, Copenhagen. EEA, 2013. Managing Municipal Solid Waste. European Environment Agency, Copenhagen. Fullerton, D., Kinnaman, T.C., 1995. Garbage, recycling and illicit burning or dumping. J. Environ. Econ. Manage. 29, 78–91. Garoupa, N., 2000. The economics of organized crime and optimal law enforcement. Econ. Inq. 38, 278–288. Gray, W.B., Shimshack, J.P., 2011. The effectiveness of environmental monitoring and enforcement: a review of the empirical evidence. Rev. Environ. Econ. Policy 5, 3–24. Grossman, H.I., 1995. Rival kleptocrats: the mafia versus the state. In: Fiorentini, G., Peltzman, S. (Eds.), The Economics of Organized Crime. Cambridge University Press and CEPR, Cambridge. Guiso, L., Sapienza, P., Zingales, L., 2006. Does culture affect economic outcomes? J. Econ. Perspect. 20, 23–48. Guiso, L., Sapienza, P., Zingales, L., 2004. The role of Social capital in financial development. Am. Econ. Rev. 94 (3), 526–556. Hage, O., Soderholm, P., 2008. An econometric analysis of regional differences in household waste collection: the case of plastic packaging waste in Sweden. Waste Manage. 28 (10), 1720–1731. Holtz-Eakin, D., Newey, W., Rosen, H.S., 1988. Estimating vector autoregressions with panel data. Econometrica 56, 1371–1395. Hettige, H., Huq, M., Pargal, S., 1996. Determinants of pollution abatement in developing countries: evidence from South and Southeast Asia. World Dev. 24, 1891–1904. Ino, H., 2011. Optimal environmental policy for waste disposal and recycling when firms are not compliant. J. Environ. Econ. Manage. 62, 290–308. Isely, P., Lowen, A., 2007. Price and substitution in residential solid waste. Contemp. Econ. Policy 25, 433–443. Jenkins, R., Maguire, K., Morgan, C., 2004. Host community compensation and municipal solid waste landfills. Land Econ. 80, 513–528. Johnstone, N., Labonne, J., 2004. Generation of household solid waste in OECD countries. An empirical analysis using macroeconomic data. Land Econ. 80 (4), 529–538. Kinnaman, T., Fullerton, D., 2000. Garbage and recycling with endogenous local policy. J. Urban Econ. 48, 419–442. Legambiente, 2010. Rapporto Ecomafie 2010. Edizioni Ambiente, Milano. Marselli, R., Vannini, M., 1997. Estimating a crime equation in the presence of organized crime: evidence from Italy. Int. Rev. Law Econ. 17, 89–113. Marselli, R., Vannini, M., 2000. Quanto incide la disoccupazione sui tassi di criminalità? Riv. Polit. Econ. 10–11, 273–299. Massari, M., 2004. Ecomafias and waste entrepreneurs in the Italian market. Paper presented at the 6th Cross-border Crime Colloquium held in Berlin, Germany, September 2004. Mazzanti, M., Montini, A., Nicolli, F., 2011. Embedding landfill diversion in economic geographical and policy settings. Appl. Econ. 43 (24), 3299–3311. Mazzanti, M., Zoboli, R., 2009. Municipal waste Kuznets curves: evidence on socio-economic drivers and policy effectiveness from the EU. Environ. Resour. Econ. 44, 203–230. Mazzanti, M., Zoboli, R., 2013. International waste trade: impacts and drivers. In: D’Amato, A., Mazzanti, M., Montini, A. (Eds.), Waste Management in Spatial Environments. Routledge Studies in Ecological Economics. Mazzanti, M., Montini, A., 2009. Waste and Environmental Policy. Routledge, London. Mazzanti, M., Montini, A., 2014. Waste management beyond the Italian North–South Divide: Spatial Analyses of Geographical Economic and Institutional dimensions. In: Kinnaman, T., Takeuchi, K. (Eds.), Handbook on waste management. Edward Elgar. Mazzanti, M., Montini, A., Zoboli, R., 2008. Municipal waste generation and socioeconomic drivers: evidence from comparing northern and southern Italy. J. Environ. Dev. 17, 51–69. Ng, Y.-K., Wang, J., 1993. Relative income, aspiration, environmental quality, individual and political myopia. Math. Soc. Sci. 26, 3–23. Pasotti, E., 2010. Sorting through the trash: the waste management crisis in Southern Italy. S. Eur. Soc. Polit. 15, 289–307. Polinsky, A.M., Shavell, S., 2000. The economic theory of public enforcement of law. J. Econ. Lit. 38, 45–76. Putnam, R., 1993. Making Democracy Work: Civic Traditions in Modern Italy. Princeton University Press. Putnam, R., 2001. Bowling Alone. Simon and Schuster. Pearce, D.W., Brisson, I., 1995. The economics of waste management. In: Hester, R.E., Harrison R.M. (Eds.), Waste Treatment and Disposal. The Royal Society of Chemistry, London, Cambridge. Shavell, S., 1997. The optimal level of corporate liability given the limited ability of corporations to penalize their employees. Int. Rev. Law Econ. 17, 203–213. Shinkuma, T., 2003. On the second-best policy of household’s waste recycling. Environ. Resour. Econ. 24, 77–95. Shinkuma, T., Managi, S., 2011. Waste and Recycling: Theory and Empirics. Routledge, New York, USA. Sullivan, A.M., 1987. Policy options for toxics disposal: Laissez-faire, subsidization, and enforcement. J. Environ. Econ. Manage. 14, 58–71. Tabellini, G., 2010. Culture and institutions: economic development in the regions of Europe. J. Eur. Econ. Assoc. 8, 677–716. Wooldridge, J.M., 2001. Econometric Analysis of Cross Section and Panel Data. The MIT Press, Cambridge, MA.