Environmental Compliance, Governance and Deforestation in Developing Countries: Theory and Evidence

Transcription

1 Environmental Compliance, Governance and Deforestation in Developing Countries: Theory and Evidence Gaoussou DIARRA Sébastien MARCHAND Centre d Etudes et de Recherches sur le Développement International (CERDI) This draft: November 2010 Abstract This paper analyses mechanisms underlying firm s compliance with environmental regulations. By using a political economy approach through a Principal-Agent model, we examine how the dynamic of environmental compliance could affect governance infrastructures by shaping the structure of institutional incentives related to stakeholder s behaviors in regulation process. Taking the case of deforestation issues in developing countries, we show that harvester s incentives to comply with forest policies and efficiency of regulatory institutions are strongly in an interaction process as firms are incited to bypass rules by using bribe or others corruption tools. In turn, government s incentives to design sustainable regulatory norms and efficient regulatory institutions are influenced in one hand by firm s regulatory compliance, and in other hand by relative instability in the political framework notably through political corruption by lobbies. Empirical estimations, using cross section data for developing countries, are supporting main theoretical predictions. After having design a new indicator of petty corruption as well as a new indicator of environmental compliance, we find that more compliance decreases petty corruption while this effect is strengthened by judicial efficiency. Moreover, we show that environmental compliance affects positively judicial efficiency and reduces deforestation while this last effect is more likely in judicial efficient countries. These results highlight that stakeholders compliance (demand for good governance) is crucial in the process of sustainable institutional reforms and environmental performances. Keywords: Corruption, Environmental compliance, Deforestation, Political Economy, Governance. JEL codes: D73, Q23. Ph.D candidate in Development Economics and International Relations. Contact: Gaoussou.Diarra@u-clermont1.fr. Corresponding author: Ph.D candidate in Development Economics. Contact: Sebastien.Marchand@u-clermont1.fr, Tel: (+33) , Fax: (+33) We are grateful to Vianney Dequiedt, Eva Trujillo of the World Economic Forum for her useful help for providing data and participants to the CERDI doctoral seminar and to LEO (Orléans) science development seminar. We also thank the multi-training program International Governance, universalism and relativism rules and institutions: What roles for international institutions? for financial support. 1

2 (S)ustainable development depends upon good governance, good governance depends upon the rule of law, and the rule of law depends upon effective compliance. None are sufficient alone, but together they form an indivisible force that is essential for survival and for sustainable development. Zaelke et al. (2005, p.29-30) 1 Introduction Moving onto the path of sustainable development requires that human behavior change to address major environmental issues such as biodiversity loss and climate change. In this context, understanding factors of deforestation is relevant to improve sustainable environment for various reasons. For instance current global warming issue highlights the importance to reduce emissions from deforestation, forest degradations, and enhance forest carbon stocks in developing countries (REDD+) To respond to these issues, governance systems have to become more effective as it describes the way available to shape human actions towards its common goals as sustainable development. In this system, institutions, conceived as rules of the game (North, 1994) provide practices and guide economic, political and social human interactions. As noticed by (Zaelke et al., 2005, p.38), all forces that can influence human behavior are potential tools of governance. In this context, good governance depends on rule of law which refers to a set of rules applied fairly, efficiently and predictably by independent institutions in a framework of public-private interaction process. Thus, a set of social norms comprising, rule of law, anti-corruption, and accountability have been put forward to enhance governance. In this context, compliance is a substantial element of good governance. Compliance with the law is related to the respect and the enforcement of legal system. As a result, compliance may be treated as an indivisible part of rule of law: rule of law cannot have meaning without compliance. The growing focus on compliance seems to be important to enhance the success of efforts to strengthen rule of law, which in turn will improve the achievement of efforts to promote good governance, and thereby allow the society to reach sustainable development. Indeed, in most studies on governance, emphasis is put on the supply side of reforms. The importance of taking into account the demand side of good governance reforms imply to be aware of factors driving stakeholder s compliance (Odugbemi and Jacobson, 2008; Young, 1979). This paper aims at bridging this gap by analyzing relationships between governance infrastructures and patterns of keys stakeholders compliance with regulatory policies, notably in environmental field such as deforestation. The importance of compliance is nowhere more important than in the field of environment and sustainable development. For instance, in a poor legal and judicial system, in which non-compliance and corruption become pervasive and law are not enforced, the respect for the rule of law is lessened, environmental degradation is more likely to occur, and progresses towards sustainable development are weakened. In this context, improvements of governance infrastructures and more particularly norms enforcement and compliance have been widely recognized. Thus, factors underlying actors compliance concerned with regulatory policies allow to understand corruption phenomenon and functioning of governance infrastructures. Moreover literature on regulatory compliance has focused on whether a firm complies with existing regulations, and on effects of enforcement on a firm s compliance behavior (Magat and Viscusi, 1990; Deily and Gray, 1991; Laplante and Rilstone, 1996). Nonetheless, in these studies the role of bribery and other political economy as- 2

3 pects of enforcement and compliance were under-investigated. While compliance with regulations was treated in literature in terms of compliance versus non-compliance (Deily and Gray, 1991), compliance level (Laplante and Rilstone, 1996) or duration of non-compliance episodes (Nadeau, 1997), studies on corruption have been focused on analysis of factors dealing with selection and incentives of actors taking policy decisions in a State with some interests on bureaucratic or political institutions structures. Besides, some evidences are also found about over-compliance of firms in manufacturing sector. Then, many firms are likely to comply with environmental regulations even when these regulations are weak or non-existent, well known as the Harrington paradox (Harrington, 1988). Analyzing sustainability of governance reforms implies to know why is there too few demand for good governance in some economies, or, to understand the persistence of some bad institutions such as those self-sustaining corruption. For instance bureaucratic corruption can be attributed to a lack of sufficient political will while political corruption can be attributed to lack of adequate political competition. As good governance reforms are gradual processes and necessitate investment in legal and administrative infrastructures, political instability is likely to create an environment under which corruption becomes more pervasive and tends to persist (Damania et al., 2004; Mishra, 2006). Therefore, governments will be constrained in its abilities to enforce compliance with chosen policies while being vulnerable to lobbying activities. Hence corruption is conceived along different dimensions which are interlinked: grand versus petty corruption. Grand corruption which involves higher decision-levels in a country, is conceptually quite different from petty corruption. It includes cases when politicians or high-ranking civil servants manipulate a country s management or regulation of infrastructure industries to gain exclusive benefits (Knack, 2006; Kenny and Soreide, 2008). A main contrast between petty and grand corruption is related to monitoring mechanisms, which can be far better organized when petty corruption is the main issue. With grand corruption, elected politicians are often directly or indirectly controlling supposed monitoring mechanisms, including media as well as judicial system. In this perspective, the fundamental rationale for giving special attention to corrupt actions in forestry is that those actions undermine the government s apparatus. This is critical in a sector such as forestry that generates many social and environmental impacts that call for a strong and clean intervention from State. If government is corrupt, there is little hope that illegal operations by other actors in the economy will ever be controlled (Contreras-Hermosilla, 2000). Anti-corruption policies limited to forestry are unlikely to work in countries with high corruption levels, which require systemic institutional changes. Corruption may matter for deforestation through design and implementation of land use plans. Indeed, grand corruption is likely to influence land use planning because decisions are made at high levels of government and large sums of money (or political support) are required to manipulate involved people. Petty corruption is likely to occur when local officials allow illegal encroachment on forest areas (Angelsen et al., 2009). Strategy for combating forest crime rests on an assessment of governance situation of the country, the streamlining of policy framework, and the focus on operational activities around a prevention, detection and suppression framework. However, since forestry does not operate in complete isolation from other sectors, enhancing compliance with forest law will depend on actions by judiciary and law enforcement agencies. Hence, this article examines interrelationships between governance, environmental compliance and forest cover in developing countries. 3

4 Positioned on the same lineage as Damania et al. (2004) but without being primarily interested by the channel of political instability as the way through which corruption may affect environmental compliance, we develop a theoretical model that explains how compliance affects i) governance infrastructures as judicial efficiency and petty corruption, and ii) deforestation through its interactions with governance infrastructures. More precisely, we use a two-stage model where a profit-maximizing representative harvester selects the optimal amount of harvested land to be cleared given the available technology in forestry. I the second step the logging firm is inspected by a bureaucrat (inspector) and she might be willing to bribe this inspector in case of non-compliance. Firm-inspector interactions are shaped by judicial efficiency and authorized harvest quota previously set. Indeed, the first step presents a model of policy-maker corruption, following Grossman and Helpman (1994). Policy maker designs the forest policy i.e the maximum harvest intensity and legal policy i.e judicial efficiency. The incumbent government has to draw both policies but the logger acts as a lobby so that policy-maker maximizes a weighted sum of social welfare and received bribes. Consequently the threshold level of harvest quota and judicial efficiency are influenced by the harvester s lobbying. We find fourth main theoretical results related to the effects of political instability on deforestation and to environmental compliance on deforestation and governance. First, we display that political instability i) strengthens deforestation via grand corruption (firm s lobbying) and ii) has no effect on deforested areas whether judicial efficiency is high. Second, concerning the effects of environmental compliance on environmental performances and governance, we highlight tree main results. First, an increase in environmental non-compliance strengthens petty corruption (between firm and inspector) via instruments of regulatory policy and judicial efficiency. Hence we show that link between cause (non-compliance) and outcome (corruption) interacts more complexly through the quality of legal and regulatory system which monitors agent s economic interactions and enforces compliance with law. Second, we find that an increase in harvester s non-compliance results in a decrease in deforestation via the previous level of judicial efficiency. This result shows the importance of rule of law and the quality of legal system to shape agent s incentives to comply with law and thus to challenge environmental issues. Third, we highlight that an increase in environmental compliance leads to an improvement of judicial efficiency in the next period. This outcome suggests that a way of improving rule of law is to enhance environmental compliance. Put differently, a solution to promote sustainability of reform consists in influencing agents economic behaviors by creating social norms allowing people to believe that the environment is threatened, that their actions may help reduce that threat, and that they have an obligation to do so in order to develop an environmental consciousness (Zaelke et al., 2005, p.41). Moreover, we test theoretical predictions using a dataset of developing countries after having design a new indicator of petty corruption as well as a new indicator of environmental compliance. Our empirical estimations support main theoretical predictions. First, we find that (i) political instability affects positively deforestation in more corrupted countries and less judicial efficient ones and (ii) that environmental compliance seems to reduce deforestation but only in countries where legal system is efficient to monitor economic activities and challenge disputes. Second, we show that more compliance decreases petty corruption while this effect is strengthened by judicial efficiency. However we find that the efficiency of legal system increases significantly petty corruption whether compliance is low, suggesting that only judiciary reforms are not enough to overcome corruption issues if people are not 4

5 incited to comply with rules. This argumentation is also validated where compliance affects positively judicial efficiency. The paper is organized as follows. Section 2 presents a political economy model while Section 3 gives the econometric approach and data used. Section 4 introduces econometric results and section 5 concludes and discusses policy implications. 2 Baseline Model A three-tier hierarchy agency model is used: firm-harvester/inspector/government 1. We assume that all parties are risk neutral. The representative harvester has right to cut timber in a given public forested land (like a concession). The harvester decides the volume to harvest knowing that an inspector will monitor his production. We first define the time sequence of actions and specify our model. Then firm-inspector interactions are studied to determine harvesting and reporting decisions as well as optimal bribe. Finally, policy decisions are investigated. 2.1 Time Sequences of Decision The model is a sequential finite-period stage game. The timing of events is given below: At date t = 0, the harvester determine its political contributions offered to the incumbent government. The level of these contributions is linked to environmental (here the forest policy) and legal policies. At date t = 1, the incumbent government selects the forest and legal policies to maximize his pay-off. At date t = 2, the harvester determines his harvest and reported effort. At date t = 3, the inspector controls firm s production and an optimal level of bribe is determined with a probability to be discover according to the quality judicial efficiency i.e the quality of the audit agency. At date t = 4, the political election begins and the incumbent government could be unseated by a rival with a given probability. At date t = 5, after election, the lobbying process starts again between the new government and harvester. A backward induction process is used for the resolution of the game. Hence the forest and legal policy is given for the harvester and inspector in the second stage. 1 An audit agency takes place in the game but only to monitor inspector s reported harvest production. Hence the audit agency does not take part into the equilibrium process. 5

6 2.2 Firm-Inspector Interaction The policy-maker designs a forest policy in order to reduce deforestation, to preserve biodiversity or to implement a durable forestry management. This policy consists to set the maximum harvest effort, e required i.e a harvest quota. We assume that the harvester tends to produce more than the quota and to set e > e implying a positive level of non-compliance v = (e e). Given that harvest production results in environmental damages, the government has to control harvest effort and imposes a bureaucrat (for instance an inspector) who monitors the firm. Moreover the harvester has a private information (asymmetrical information) on his harvest production so that the principal has to use an inspector to monitor firm s reported information. However, the firm may propose a bribe B to inspector in the case he does not comply with harvest quota. This corrupt behaviour is defined as petty corruption because it occurs between the firm and inspector 2. Then the harvester s bribe incites inspector to declare that quota is respected. In the model, inspection is not probabilized because we suppose that the harvester is alone which makes obvious the inspection. In other words, given that the inspector is risk-neutral, he could accept or refuse the bribe if the amount of this contribution is not enough to optimize his pay-off. If the inspector accepts bribe, he reports harvest effort e e. Moreover, we assume that the inspector receives from the regulator (government) a fixed wage (w) and a variable wage (R). This last component of inspector s remuneration, R, is considered as a reward for reporting the true harvester s effort. Then R depends positively on the level of non compliance (v, R/ v > 0) but as public resources are scarce, R should not increase indefinitely with v. This award is a positive incentive used as a regulatory tool by the incumbent government to persuade the inspector to comply with his original purpose i.e to report the true harvest production. However, knowing that the inspector could be bribed by harvester, the regulator implements an audit of the harvest effort. The audit uncovers the true harvest effort with a probability λ (0, 1). Therefore, the level of λ could be an indicator of the efficiency of auditing process as noticed by Damania et al. (2004). If e > e, a fine f(v) is imposed to the harvester (f H ) and inspector (f I ). Thus, the fine for corruption depends positively on non-compliance f(v)( f v > 0). Hence, the harvest effort will depend on the forest policy (e) and legal policy (λ), built for penalties in the case of non compliance. Thus, harvest effort will be e = e(e, λ) in the cheat/bribe strategy and e S = e(e, λ) in the safe strategy. Let H(e) be the net harvest function of the harvester from its harvest effort, with the land-holding cost and timber prices integrated. We assume that this logger is price taker given that he trades on international market. The net harvest function depends on the logging effort (which is the logger s choice variable), and takes the form H(e), with standard properties: H e > 0, H ee < 0. The net harvest function is H(e S ) in the safe case and H(e) in the cheat strategy. Consequently, the strategies and their related pay-off are given in the following table 1. 2 In contrast, corruption occurring between the harvester and incumbent government is considered as grand corruption. 6

7 Table 1: Strategies pay-off Strategy Harvester Inspector Safe H(e S ) w + R(v) Cheat H(e) [B + λf H (v)] w + B λ[f I (v) + R(v)] Ψ H (e, e) Ψ I (e, e) Net = H(e) [B + λf H (v)] H(e s ) = w + B λf I = (v) + C(v)] w R(v) = B (1 + λ)r(v) λf I (v) Following Damania et al. (2004) and Wilson and Damania (2005), we begin by determining the optimal level of bribe and effort intensity. The process of this game can be divided in two stages. First, the logger sets the effort level to maximize the joint pay-off (Ψ J Ψ H +Ψ I ). Then, the bureaucrat and logger share the surplus through a Nash equilibrium process. Therefore, taking the policy parameters as given, the level of reported and actual effort are chosen to max e,e ΨJ Ψ H + Ψ I = [H(e) [B + λf H (v)] H(e s )] + [B (1 + λ)c(v) λf I (v)] (1) The first order conditions are Ψ J / e = H e λf H e λf I e = 0 (2a) Ψ J / e = λf H e + λf I e (1 + λ)c e = 0 (2b) Note that equation (2a) specifies that in equilibrium, the actual effort is set such that the marginal revenue from production equals the marginal expected penalty. By equation (2b), the marginal cost of compliance (for the logger and inspector) is equated to the marginal expected cost of non-compliance (i.e. the expected marginal fine). The equilibrium bribe is determined by a Nash bargain between the harvester and bureaucrat. In this process, both parties share equally the benefit of the bribe strategy (non compliance) i.e they have the same bargaining power. To be successful, the bargaining has to respect the reservation values of the bureaucrat and logger. By the harvester s net pay off (table 1, col.1), the bribe is paid by the firm i.e the bribe strategy is dominant if B < H(e) H(e S ) λf H (v) B. By the inspector s net pay off (table 1, col.2), the inspector accepts the bribe if B > (1 + λ)r(v) + λf I (v) B. Therefore, the optimal bribe is determined by the following Nash bargain The first order condition is max B (ΨH Ψ I ) (3) (Ψ H Ψ I )/ B = [H(e) H(e S ) + (1 λ)r(v) λ(f H f I ) 2B] = 0 (4) Hence, the optimal bribe is B = 1/2[H(e) H(e S ) + (1 λ)r(v) λ(f H f I )] (5) 7

8 The optimal bribe resulting of this process has to lie between [B, B] so that the global benefit of the bribe strategy is higher than the global loss of this strategy. The comparative static properties of the equilibrium show that the bribe raises if the inspector s emoluments (rewards) increase (Mookherjee and Png, 1995). Intuitively, the opportunity cost of cheating for the inspector raises, so the inspector demands and receives a bigger bribe. Besides, to ensure that higher fines (f) reduces the optimal level of bribe (B ), the term f H f I must be positive. This assumption implies that the fine paid by the recipient (inspector) is lower than that paid by bribe giver s fine (harvester) (Damania et al., 2004; Delacote, 2008). The intuition is as follows: if the regulator punishes more severely bribe-takers than bribe-givers, the inspector would incite the harvester to increase the bribe to compensate for the inspector s expected fine. Hence, as suggested by Mookherjee and Png (1995, p.151), one way to reduce the bribe is to raise the penalty [...] on the bribe-giver (factory), while reducing the penalty [...] on the bribe-taker (inspector). The comparative static properties of the equilibrium highlight also that the rule of law influences negatively petty corruption ( B/ λ < 0). Given that, we have the following proposition concerning the influence of environmental non compliance. Proposition 1 An increase in environmental non-compliance strengthens petty corruption but this effect is indirect via instruments of regulatory policy (R and f) and depends on judicial efficiency (λ). Thus, environmental non-compliance has an indirect effect on corruption but this effect depends on i) positive (R) and negative (f) incentives provided by the incumbent government to prevent corrupt behaviours of the inspector and ii) judicial efficiency (λ) (see proof A.1). When the quality of judicial efficiency is low, the effects of environmental non-compliance through bureaucrat s reward on petty corruption is reinforced whereas the effects through fines is weaker. The intuition is as follows: in a country with low judicial efficiency, more rewards for the inspector, provided by government to reduce corruption between the harvester and inspector, surprisingly raises these unauthorized interactions. The idea leans on the inspector s weight in negotiations which increases with the level of rewards received. The inspector could ask a more important bribe to compensate rewards knowing that the probability of detection of the cheat is very low. Thus, in country with a weak rule of law, a rise of inspector s rewards seems not to be a good solution to reduce petty corruption. However, when the quality of the legal framework is high, this leads to a weakness of the relationship between the environmental non-compliance, bureaucrat s emoluments and petty corruption. The intuition is as follows: knowing that the detection of cheat is nearly certain, the inspector prefers to keep his reward and not be bribed by the firm. Moreover, the effect trough negative incentives (fine) depends on elasticity of fine paid by each agent (the harvester and inspector) to non-compliance. Above, we suggest that firm s fine is higher than inspector s one. Hence in this case a marginal increase of the fine for non-compliance is higher for the firm than for the inspector. In this case, the model suggests that a high level of firm s non-compliance could be associated with few firm s petty corruption. The intuition relies on the expected fine paid by the firm which increases the cost of the cheat strategy implying that the harvester prefers to not bribe the inspector. More precisely, the firm knows that his non-compliance behaviour will likely be detected with or without bribe, as judicial efficiency is high, 8

9 and so will prefer to spare the amount of bribe. Thus, in country with efficient judicial institutions, this leads to reduce opportunities of petty corruption. In contrast, when elasticity of inspector s fine to non-compliance is higher than which of harvester, non-compliance increases petty corruption. As expected, the inspector asks a more important bribe because he knows that the detection is more likely to occur (opportunity costs of cheating raises). Finally, from this first proposition, it is worthy noting that public policy aiming at reduce petty corruption in environmental field should consider underlying incentives which influence firms behaviours in term of environmental compliance. 2.3 Policy Determination We now turn to the political process after having described the harvester-inspector interactions process. Remind that we proceed backward so that the political process, which determines the forest policy and legal policy, takes place before harvester-bureaucrat outcomes. The incumbent government has to draw policy but the logger act as a lobby. Consequently, the threshold level of harvest quota (e) and quality of judicial efficiency (λ) are influenced by the harvester s lobbying. Following Damania et al. (2004), we assume a different timing in the policy determination between these two policies. Forest policy could be changed instantaneously whereas effects of legal reforms (to enhance judicial efficiency) are more longer. This slow legal process is captured as follows : a change in enforcement expenditures, initiated in one period (one political period), has an impact on environmental compliance in the following period. Thus, the incumbent government s investment decision will influence the future governments policy-making. As Damania et al. (2004), there are costs to implement legal policies defined as enforcement costs of regulations C. Those cost depends on the quality of judicial efficiency λ, determined in previous period, and collected fines f(v) = f H (v)+f I (v). C τ ( λ τ 1 ) Cτ Thus we have, C τ (λ τ 1, f τ (v t )) with > 0 and ( f τ ) > 0. However, we assume that judicial efficiency depends on the previous level of non-compliance arguing that the incumbent government could decide to implement a policy in order to improve efficiency of legal framework i.e λ τ (v τ 1 ). Following Grossman and Helpman (1994), the incumbent government i receives political contributions S defined as grand corruption from the logger. The government derives utility from political contributions and also from the welfare of voters (W ). The incumbent government draws the level of policy parameters in order to maximise its pay-off considering S as given. Consequently, the current utility of the policy-maker is a weighted sum of S and W : U i τ = S i τ + α i W i τ Where α i measures the weight of social welfare in government s utility. Social welfare represents the sum of utility of all agents in the economy in period τ (τ = 0,..., T ). W i τ = H(e τ ) D(e τ ) C τ (λ τ 1 (v τ 2 ), f τ (v)) Social welfare represents the consumer surplus, harvest profit, environmental damages of harvest production (D(e)) and cost of enforcing policies which are the forest policy, legal policy and inspector s 9

10 payments (w and R) 3. To model political competitions and so political instability, we assume that a political actor j challenges the current government i. The political competition occurs after the firm-inspector interaction and so after the forest-legal policies determination. We assume that i stays in power with probability ρ and loses the political contest with probability (1 ρ). After elections, the political winner has to implement the forest and legal policies. Thus, the harvester gives new political contributions. Therefore, given that political instability is known by the harvester and government, they will take it into account while making their optimal choices. Suppose that we have T = 2 and δ, the discount factor. Thus, the discounted expected pay-off of the incumbent government (i) is And, for challenger j, the expected pay-off is U i = U i 0 + U i 1 = S i 0 + α i W i 0 + δρ[s i 1 + α i W i 1] U j = δ(1 ρ)[s j 1 + αj W j 1 ] Given that, the effects of legal policy occur in next period, so in τ = 1 this policy is implemented and in τ = 2, its effects take place. Following Grossman and Helpman (1994) and Dixit et al. (1997), the equilibrium in a common agency model maximises the joint surplus of all parties. From lemma 2 of Bernheim and Whinston (1986), the policy vector a i = (ei τ, λ i τ ) must satisfy two necessary conditions, which are CI : a i arg max U i τ (6a) CII : a i arg max E(Π) + U i τ (6b) E(Π) = Π i τ + δ[ρ(π i τ+1 ) + (1 ρ)πj τ+1 ] is the expected profit of the harvester. The expected profit Π of the logger when government i is in power (in τ) is Π i τ = H(e i τ ) B i τ λ τ 1 (v τ 2 ) f H (v i τ ) S i τ Consequently, Π i τ+1 (Πj τ+1 ) is the expected profit when the incumbent i (rival j) wins power in τ + 1 which is Π τ+1 = Π i τ+1 + Π j τ+1 = δρ[h(e i τ+1) Bτ+1 i λ τ (v τ 1 ) f H (vτ+1) i Sτ+1] i + δ(1 ρ)[h(e j τ+1 ) Bj τ+1 λ τ (v τ 1 ) f H (v j τ+1 ) Sj τ+1 ] Condition 6a states that the incumbent government determines the forest policy as well as legal 3 Inspector s payments are dropped from the social welfare because there are paid by the incumbent government and received by the inspector. 10

11 policy to maximize its own welfare (U i ), given the offered contribution schedules (S i ). Condition 6b denotes that the equilibrium of the forest and legal policies maximizes the joint welfare of harvester and government. The optimal policy vector a τ maximizes i) CI (equation 6a) and ii) CII (equation 6b). Concerning the equilibrium of harvest quota e, the first order conditions satisfy S i τ e i τ [ H i + α τ e i τ Thus, e is implicitly given by: S i τ e i τ Di τ e i τ [ H i + α τ e i τ ] + Hi τ e i τ Di τ e i τ ] = 0 λ τ 1 (v τ 2 ) f H τ e i τ Si τ e i τ = 0 S i τ e i τ = Hi τ e i τ λ τ 1 (v τ 2 ) f H τ e i τ (7) S i τ λ i τ Concerning the equilibrium of legal policy λ, the first order conditions satisfy Si τ λ i τ + δ [ ρ S i τ λ i τ ( ατ i Cτ+1 i λ i πi τ+1 τ λ i τ [ ( + δ ρ ατ i Cτ i ) ( λ i + (1 ρ) α i C i )] τ τ τ λ i = 0 τ λ i τ vτ 1 i fτ+1 H ) ( + (1 ρ) ατ i Cτ+1 i λ i πj τ+1 τ λ i τ λ i τ vτ 1 i fτ+1 H )] = 0 Thus, λ is implicitly given by: S i τ λ i τ ( π i = δρ τ+1 λ i τ λ i τ vτ 1 i fτ+1 H ) ( π j τ+1 + (1 ρ) λ i τ λ i τ vτ 1 i fτ+1 H ) (8) These two equations defined the political equilibrium in the model. Equation 7 stands that the marginal profit of harvesting activities equals to the marginal cost divided between costs of noncompliance and political contributions. equals the marginal profit in τ + 1. Equation 8 states that the marginal cost of legal policies Having defined the political equilibrium of the model, we now turn to examine interrelationships between forest policy, environmental regulation and governance infrastructures. We investigate influence of political instability and environmental compliance on both policies i.e legal one (rule of law) and forest policies. First of all, comparative static properties of the equilibrium show that grand corruption increases the forested quota i.e the forest policy becomes more flexible ( e S > 0). This result is consistent with conclusions emerging from the literature (Barbier et al., 2005; Amacher, 2006; Delacote, 2008). The intuition is as follows: corruption is defined like an informal pro-deforestation policy enhancing rent seeking activities. Then, corruption allows a miss-management forestry production which leads to an over-extraction of resources to enhance the rent provided by forestry activities. Also, a well-developed corruption can enhance illegal activities which are often seen as pro-deforestation. 11

12 Second, comparative static properties of the equilibrium states that political instability decreases judicial efficiency i.e regulatory policy becomes weaker ( λ ρ < 0). This result is consistent with conclusions emerging from the literature. As noticed by Damania et al. (2004, p.372), unstable regimes alters the firm s lobbing incentives [...] (and) the incumbent government s willingness to modify policies in response to political contributions. When policies are uncertain, an inefficient regulatory structure allows firms to evade future regulations through bribery. Thus, in unstable regimes, formal policy settings may bear little relation to the real effects of policies. Given that, we have the following proposition concerning the influence of political instability on deforestation through grand corruption. Proposition 2 An increase in political instability strengthens deforestation but this effect is indirect via grand corruption (firm s lobbying) and judicial efficiency (see proof A.2). An increase in political instability i) reduces judicial efficiency and ii) raises opportunities of corruption and so the willingness to be corrupted. Consequently, this leads to more miss-management forestry activities and so increases deforestation. Moreover, positive effects of instability on corruption is conditioned to judicial efficiency. Hence a better judiciary environment reduces positive effects of instability on corruption. Thus, deforestation is less influenced by the effects of corruption and unstable regime when judicial efficiency is high. We now turn to the following proposition on environmental compliance and deforestation. highlight two propositions on the role of environmental non-compliance on deforestation belonging to the dynamic process of compliance and judicial reforms. Proposition 3 A high judicial efficiency weakens harmful effects of the current harvester s environmental non-compliance on current deforestation (see proof A.3.1). If the judicial efficiency is significant, current environmental non-compliance will be revealed so that the government could implement a more stringent forest policy in response to this high environmental non-compliance. However, if the quality of the legal framework in challenging regulation is low, the incumbent government will not be incited to take into account harvester s environmental noncompliance due to a lack of institutional infrastructures or a lack of political willingness to reduce environmental damages. Proposition 4 An increase of past environmental non-compliance strengthens the current deforestation but this effect occurs via the past level of judicial efficiency (see proof A.3.2). The intuition for this result is that if environmental non-compliance is significant, any judicial reforms will not necessarily lead to a reduction of deforestation given that the demand side of governance is not taken into account. 4. Hence good governance reforms are likely to be unsustainable and inefficient because firms compliance is low. However in the case where stakeholder s compliance with 4 Put differentially, despite an improvement of judicial efficiency, stakeholders does not increase their compliance with the law. This assumption relies on the fact that compliant behaviors are more due to factors not linked with supply of good governance. We 12

13 environmental rules is high, the effect of judicial efficiency on deforestation becomes non significant. This last result states that in a such economy, the supply of good governance in terms of legal system reforms (or judicial efficiency) does not matter for the pattern of deforestation. Proposition 5 An increase in environmental compliance leads to improvements of judicial efficiency in next period (see proof A.4). This result states that an improvement of environmental compliance improves the quality of governance infrastructures and more particularly the efficiency of legal framework. In other words, policy recommendations should be focused on incentives of compliance in order to increase the quality of environmental regulations, a condition to enhance the legal framework and to reduce environmental damages. 3 Econometric Approach and Data Used 3.1 Econometric Specification The theoretical model highlighted interrelationships between environment regulations, governance infrastructures and environmental damages. The econometric approach consists in testing whether: (i) environmental non-compliance strengthens petty corruption via judicial efficiency (Proposition 1) (ii) political instability strengthens deforestation via grand corruption (Proposition 2); (iii) current environmental non-compliance decreases deforestation via the past level of judicial efficiency (Proposition 3) 5 ; (iv) environmental compliance improves judicial efficiency in next period (Proposition 5). To test our model, a cross-country data is used with a three-equation econometric model of petty corruption (equation 9), deforestation (equations 10 and 11) and judicial efficiency (equation 12). Petty Corruption Equation To test proposition 1, we run the following regression: P etty i = Compliance i + 2 Judicial i + 3 Compliance i Judicial i + k k W i + υ i (9) where Petty i is the level of petty corruption in the country i (averaged on ), Compliance i is environmental compliance (averaged on ) and Judicial i is the quality of the legal framework (averaged on ). W i are control variables, are coefficients vector to estimate and υ i are the error term. Following literature, we divide factors i.e W i of petty corruption into four main categories: (i) economic factors, (ii) political and legal determinants, (iii) social structure and (iv) geographical features. First, on economic factors, Log GDP per capita is introduced to control for development process which lessens the level of corruption. Trade openness is used because it reduces corruption by depressing the presence of potential rents (Ades and Tella, 1999). A variable of rents from Oil is also used to control for the presence of potential rents which conditioned the emergence of petty corruption. 5 Having no data on past environmental compliance we can not test proposition 4. 13

14 Second, on legal factors, Treisman (2000) and Damania et al. (2004) argue that democratic regime lessens corruption. Thus a Democracy variable, measuring the extent to which there is a societal consensus supporting democratic principles, are introduced. Legal origins of the law and regulations are introduced to control for differences between major family of law. La Porta et al. (1999) show that English common law countries have less-corrupt societies than French civil law ones because a civil law country is often associated with an important government ownership and regulation which are associated with many adverse impacts on markets such as corruption. Third, we include Latitude because temperate zones with healthier climate and more agricultural productivity could develop their economic activities and their institutions and so lessens the potential opportunity of corruption (La Porta et al., 1999). Deforestation Equation To test proposition 2, we run the following regression: Deforest i = α 0 + α 1 Instab i + α 2 Corrupt i + α 3 Instab i Corrupt i + k α k Z i + ε i (10) where Deforest i is the growing forest stock in the country i (between 2000 and 2005), Instab i is the political instability (averaged on ), Corrupt i is the degree of grand corruption (between 2000 and 2007) and Judicial i is the quality of the legal framework (averaged on ). Z i are control variables, α are coefficients vector to estimate and ε i are the error term. To test proposition 3, we run the following regression: Deforest i = α 0 + α 1 Compliance i + α 2 P astjudicial i + α 3 Compliance i P astjudicial i + k α k Z i + ε i (11) where Deforest i is the growing forest stock in the country between 2000 and 2005 i, P astjudicial i is the quality of the legal framework in the previous period ( ), Compliance i is the level of environmental compliance (averaged on ), Z i are control variables, α are coefficients vector to estimate and ε i are the error term. Concerning control variables, literature on factors of deforestation is substantial. First, the path of deforestation is often associated positively with the initial forest area (Forest Lag) for three main reasons: i) the scarcity effect implies that the remained forested lands are more preserved ii) logging or forest conversion activities are more expensive when forest areas are low; iii) the non-forested lands (in proportion to total area) are more available to other land uses such as agriculture, cattle ranching, tree plantations or urbanization when forested lands are important. Second, the Environmental Kuznets Curve (EKC) approach states that the level of environmental degradation, measured by environmental indicators as deforestation, is linked to economic development (Bhattarai and Hammig, 2001; Culas, 2007). This relation implies a non linear effect of economic growth on deforestation (GDP per capita and GDP per capita 2 ): the marginal income effect is positive for low income countries and become negative for high income economies. Third, demographical effects and more particularly population pressures are associated with more deforestation. The common explanation advances is that population growth encourages the conversion 14

15 of forest land to other uses (and almost agricultural uses) by increasing the need for arable land. The growth of Population (Pop) and the extent of rural population (Rural) are introduced to control for these pressures on forested land (Cropper and Griffiths, 1994). Fourth, the relative price of timber (Timber) is an important determinant in the trade-off between keeping forest option or clearing one. Arcand et al. (2008) propose two kinds of effect of the relative price of timber if the marginal variation is temporary or permanent. A permanent increase in this relative price decreases the discount rate (more weight for long-term and durable activities) and so reduces deforestation in their dual theoretical model. On the other hand, a temporary increase could strengthen clearing activities by pushing up the discount rate (more weight for short-term activities as clearing and after agricultural activities). Thus the clearing profit is bigger than the keeping forest activities one whether the rise of relative price of timber is considered as one period shooting. Judicial Efficiency Equation To test proposition 5, we run the following regression: Judicial i = β 0 + β 1 Compliance i + k β k X i + ϕ i (12) where Judicial i is the quality of the legal framework in the country i (averaged on ), Compliance i is environmental compliance in the country i (averaged on ), X i are control variables, β are coefficients vector to estimate and ϕ i are the error term. We divide determinants into two parts: (i) economic factors and (ii) legal ones. Log of GDP is introduced as a proxy for economic development (GDP per capita). Trade openness is used as structural differences influencing judicial efficiency as well as Democracy and Political Freedom. Following La Porta et al. (1999, 2004), legal origins of law and regulation influence judicial efficiency in a country. They argue that English common law countries have more secure property rights because judicial independence is more developed than in French civil law ones. They show that this legal feature in common law countries (see Glaeser and Shleifer (2002) for an historically and theoretically explanation of judicial independence in common law countries) represents the judge s independence to executive and legislative powers which promotes judicial efficiency. Finally, a index of checks and balances is used given that checks and balances is one of conditions of judicial independence. 3.2 Data We begin by a description of institutional variables: compliance, petty and grand corruption, judicial efficiency, political instability. Table 4, page 28 summarizes descriptive statistics and Table 2, page 25 provides definitions and sources of variables used. In the literature, we found no data on environmental compliance at country level. To challenge this issue, we build an index of environmental compliance from data of several reports provided by the World Economic Forum. Compliance is predicted values of the corporate ethic 6 of firms in a given country which has been regressed on (i) a variable measuring the burdensome for businesses to 6 Corporate ethic is the ethical behavior in interactions with public officials, politicians, and other enterprises of firms in a given country. Data are averaged on the period

16 comply with governmental administrative requirements (averaged on the period ) and on (ii) a variable related to the stringency of environmental regulation (averaged on the period ) (see table 3, page 29 for more information on these data and table 5, page 29 for rankings of 15 worst and best countries in term of compliance, a higher value of this index implies a high compliance). Concerning data on petty corruption, we also found no information on this topic. Indexes of corruption commonly used (ICRG, Transparency International, WGI) capture perceptions of the extent to which public power is exercised for private gain, including both petty and grand forms of corruption. Again, to resolve this issue we build an index of grand and petty corruption from the index provided by Transparency International (averaged on ) which focuses on corruption in the public sector seen by business people, risk analysts and the general public (from 0 (highly corrupt) to 10 (highly clean)). The estimated level of grand corruption is the predicted values of the index of corruption which has been regressed on (i) the extent of the legislative strength in a country (ICRG, average on )) and on (ii) the level of accountability (WGI, average )) 7. Petty corruption is assumed to be the part non-explained by these two explanatory variables which explain mainly grand corruption. Hence petty corruption is calculated as 8 petty = 2 corruption corruption where corruption is the corruption index of TI and corruption, the predicted variable of this index i.e grand corruption. Table 6, page 30 provides ranking of the 10 most corrupted countries and table 7, page 30 provides ranking of the 10 less corrupted countries in terms of corruption, grand corruption and petty corruption (A higher value of each of these indexes implies less corruption). As a measure of judicial efficiency, the Rule of Law index developed by Kaufmann et al. (2008) capturing perceptions of the extent to which agents have confidence in and abide by the rules of society, and in particular the quality of contract enforcement, property rights, the police, and the courts, as well as the likelihood of crime and violence is introduced (from -2.5 (low) to 2.5 (high), averaged on ). Following Damania et al. (2004), we believe that this index could be reasonably expected to approach the degree to which laws are enforced. To measure political stability, the Political Stability index developed by Kaufmann et al. (2008) is used. This index of political stability measures the likelihood that the government will be destabilized or overthrown by unconstitutional or violent means, including politically-motivated violence and terrorism (from -2.5 (low) to 2.5 (high), average ). To measure deforestation, we use the Growing Stock which is defined as the standing volume of the trees in a forest above a certain minimum size. Higher growing stock implies more standing biomass, which often translates to better forest conditions. The variable used measures the change between 2000 and 2005, in cubic meters per hectare and come from Environmental Performance Index (EPI) provided by Yale University. 7 The index of Legislative Strength is provided by ICRG (from 0 (low) to 4 (strong)) and the level of accountability is the Voice and Accountability index developed by Kaufmann et al. (2008) which represent the extent to which a country s citizens are able to participate in selecting their government, as well as freedom of expression, freedom of association, and a free media (from -2.5 (low) to 2.5 (high)). 8 The Adjusted R-squared is