Does transparency have an eeffect on political bahavior? Experimental study of lobbying.

Transcription

1 Does transparency have an eeffect on political bahavior? Experimental study of lobbying. Miloš Fišar Jiří Špalek June 9, 2017 Abstract Transparency and lobbying, two terms that have direct influences on the behavior of policymakers, politicians, and bureaucrats. Models of both transparency and lobbying exist for decades and are being researched by not only academics but non-profit international organizations and institutions that provide governments with policy recommendations. The recommendations are, however, often based on research that is complicated or almost impossible to perform or relies on self-report surveys by the public officials and policy makers. The real behavior of those actors might, therefore, be hidden. Therefore, there opens an opportunity to model these situations and bring transparency and lobbying into the laboratory or field controlled environment and investigate the behavior of involved actors using methods of experimental economics. As the experimental economics is a field of economic research that has been emerging in the main flow for las few decades, the research on transparency and lobbying is, therefore, emerging as well. There have been published studies that focused on transparency and studies that investigated lobbying, however, the number of publications on experimental research on transparent lobbying is limited. Therefore, the purpose of this article is to provide an overview of experimental research on lobbying and transparency, evaluate the findings and provide suggestions for future research. JEL-Classification: D72, C91 Keywords: lobbying, transparency, laboratory experiment Department of Public Economics, Masaryk University. milos.fisar@econ.muni.cz. Corresponding author. Department of Public Economics, Masaryk University. jiri.spalek@econ.muni.cz. 1

2 This is a preliminary draft. Please do not cite or distribute without permission of the authors. 1 Introduction Everyday life of any citizen in any country in the world is influenced by legislature and decisions made by a politician, bureaucrats, and other public servants. Their profession defines the main goal of their work - to serve the public and make decisions in the interest of others. But do they follow it or are they judging their action based on their private regarding preferences? Politics, as sometimes viewed as a struggle among a plethora of groups with different interests (Mueller, 2003), may not be willing to cooperate with any researcher. The decisions made in public sector are always complex, and the decision maker requires extensive information base to make a qualified decision. As also the outcomes of the decisions are always uncertain, the decision maker seeks an expert advice. If the advisor, the expert, is an independent entity, then there should be no constraints to base the decisions on such advice. The same might apply even when the decision makers receive the advice from different interest groups who lobby for their interest. But in the letter case, there always will be the potential another story. A story where the politician, regardless of any other information, decides in line with the preferences of one particular interest group. Lobbying, by the definition, is any form of communications by intermediaries for companies or organizations aimed at influencing public office holders in relation to public policy or administrative decisions in which their constituents have a particular interest (Côté, 2006). It opens possibilities for the public administration to base their decision on needs of different parts of the society, but on the other hand, it may negatively influence other parts of the society. However, for a long time perceived the lobbying had been as only a rent-seeking problem, as an exchange of money for the favorable political decision between a politician and an interest group (Krueger, 1974; Tullock, 1967, 1980; Snyder, 1991,... ). The opinion changes in 1990 s when it was proved that the information provided by interest group or lobbyist might be beneficial and 2

3 it improves the information base for policy makers (Ball, 1991; Lohmann, 1995; Grossman and Helpman, 2001,... ). On the other hand, lobbying also tends to bring along threat for the society when the information or monetary exchange between policy maker and lobbyist is being kept secret. The non-transparent lobbying may undermine the legitimacy of a political system and democracy (Kretschmer and Schmedes, 2010) or may lead to non-effective allocation of public funds (Campos and Giovannoni, 2007). There is also only a thin line between non-transparent lobbying and corruption (Lambsdorff, 2002). It is, therefore, important to focus on the transparency and its importance for any policy making process. Transparency became a topic both in public sector research and practical public management. The empirical research on transparency in the public office usually relies on estimations, constructions of measurements or survey data. It documents effects of the transparency but barely focuses on the behavior of policymakers under different levels of transparency. This drawback might be overcome by use methods of experimental economics as it allows the research to evaluate and observe the behavior of humans in a real time in controlled environment (Fréchette and Schotter, 2015). Our study aims to investigate transparency and lobbying in the laboratory and/or in the field using an economic experiment. We focus on the potential of the role of transparency in reducing the inefficient decisions of a politician. We developed a two-person game where the politician takes action with or without regard to lobbyist preferred outcome. The decision is then evaluated by the voter who can decide whether to punish the politician or not. We compare the change of subjects behavior when transparency (common knowledge on accepting the lobbyist point of view by the politician) is introduced. 1.1 Transparency in the experiment One of the first studies on transparency in public sector took a look at the process of amendments under opened and closed rules (Frechette et al., 2003). The opened rule is a situation when the proposal is amended a couple of times before the voting by legislators. On the other 3

4 hand, the closed rule represents situation when the proposal is brought unamended to the legislators. We can view the two rules of amendment procedures as two different levels of legislature transparency of the process. The authors found that once the process is transparent (opened) then longer delays in distributing benefits appear and the distribution of benefits is more egalitarian. Their data also revealed that the closed rule causes convergence towards minimal winning coalitions, whereas under the opened rule the supermajorities are a norm. The legislative transparency in a democratic country shall improve the functioning of parliament through voters responses in the elections. Malesky et al. (2012) however argue, that this argument does not hold in authoritarian regimes. They addressed this concern via conducting a field experiment in Vietnam, a single-party authoritarian regime. There is being observed a shift towards non-governmental institutions and therefore the governmental liability does not improve there. They find very little evidence of direct positive effects of transparency there and conclude, that to implement transparent institutions like the democratic regime is complicated. The institutions there often play different roles, even though that the institutions hold similar names as in democratic countries. Anctil et al. (2004) studied the problem of transparency from a different angle. They focused on the effects of transparent economic signals on the economic outcome as the decision makersâăź uncertainty increases. The more transparent the information base for policy or decision maker is, the more contrary the information might be. The evidence of the experiment shows that once multiple equilibria arise together with increased level of transparency, the agents tend to focus on risk-dominant equilibria. That causes cause coordination failures and it affects the economic outcome as well. The results and settings of their experiment might be also applied to the problem of transparent lobbying, where lobbyists often extend the information base of a policy maker and lobbying might cause uncertainty in the decision making then. Peisakhin and Pinto (2010) stepped over the thin line between lobbying and corruption and focused their field study on the second mentioned. They found a strong evidence for a positive anti-corruption effect of transparency even in highly divided, hierarchical and unequal society. The experiment took place in Indian slums where the access to information is usually limited, especially for the poor. The results show that simple transparent access to information improves 4

5 the public policy and their right to use basic public service. The transparent process there had a direct positive effect on well-being of the poor and disadvantaged part of the society. The well-being of citizen is important effect of transparency, but not the only one. Another key outcomes is also improvement of the trust in the government, politician or public office in general. Grimmelikhuijsen (2012) conducted a lab in field experiment focusing on the trust in government and knowledge about public processes of the citizens. In the experiment, different information was provided to the visitors of a certain website regarding levels of air pollution of municipalities in Netherlands. The findings of this study show that not only the level of transparency has a minor effect on trust in government, but more importantly, cognitive and emotional processes play a significant role. The general trust in government then also influences the trust in specific governmental institutions. That contrasts with theoretical expectation and with rhetoric and policy recommendations of well know international organizations. As the trust in the government might also be a question of cultural values. Grimmelikhuijsen (2012) followed the previous study and conducted series of experiment in two developed countries in different parts of the globe - the Netherlands and South Korea. They found similar effects in both countries. The transparency might under specific condition have a negative effect on the trustworthiness of the government. In South Korea, the transparency significantly contributes to lowering the perception of governmental competence and decreases the trust in it. The magnitude of this effect is significantly differing to the Netherlands where only the negative outcomes have an influence on the perception of the governmental competence. In South Korea, there were always observed more extreme responses. The authors, therefore, conclude, that one of the keys to the citizens trust in government is the culture. The idea of transparency in public office is that the politician or policy makers will behave differently when they actions are observed by their customers, the citizens and voters. Carey (2013) modeled the issue as a game of three politicians who vote about budget division and one voter who evaluate their actions via election. Through the experiment different levels of transparency vary. The results of his study show that transparency leads to higher public budget shares and it also foster higher reelection rate. He also found, that the transparency tends to reduce the number of minimum coalitions. 5

6 Not only the division of public budget, but preparation of legislature and other actions of policy makers are often an outcome of a bargaining process or communication behind closed doors between political parties. Agranov and Tergiman (2014) took this issue into the lab. They found that public transparent communication is being used to express fair proposals and the private channels lead to lobbying. Once the transparent communication was forced in the experiment, the lobbying activities and quid pro quos bargains nearly stopped. Kartal and Tremewan (2016) study effects that the transparency has on information transmission and decision-making. In the experiment can an advisor of decision maker accept a side payment from a third party or not. The transparency allows the decisions maker to reflect the behavior of the advisor once he makes the decision. Their theoretical and experimental findings contrary to other studies, the transparency is not harmful and it may help the quality of the decision. However, they show that the positive effects are valid only when the transparency is mandatory; when it is voluntary the results are weaker. 1.2 Lobbying in the experiment Even tough that on lobbying has been empirically researched for decades, there exists only a limited literature on lobbying in the laboratory. One of the possibilities how to take a look at the lobbying problem is interpreted as a signaling game. (Potters and van Winden, 1996) replicated theoretical assumption and empirical findings concerning equilibrium selection and costly signals. Their experiment two players where one is making the decision and the other can transmits a message to him before the action, he signals what the outcome he is looking for. The results of the experiment show, that these messages have an impact on the decision maker even though that the messages are costly. Another study by the same authors also used a signaling game to investigate lobbying (Potters and van Winden, 2000). In this study the authors conducted the experiment not only with usual experimental subjects (university students) but performed an experiment with a professional lobbyist as well. They found that part of the professionals behaved more in line with theoretical predictions and earned a significantly larger amount of money, but for most of them was found 6

7 no difference in behavior compared to students. The paper is important in answering one of the potential methodical issues of a further research on lobbying in the lab. The authors base on their findings a theoretical model of costly lobbying that was later confirmed by an empirical study as well (Helland, 2008). Helland (2008) found that the prediction that the probability of favorable outcome increases with lobbyist investment into the outcome, however, hinge on the structural characteristics of the municipality and distribution of electoral seats per voter there. Legislature process often faces many direct and indirect lobbying activities. Bergan (2009) focused in the field experiment on the indirect option. He used an campaign towards legislators in New Hampshire who received differently phrased and framed based on the districts. The results show that the grassroots campaigns have a substantial effect on the legislators and their policy decisions. Another grassroots campaign investigated (Richardson and John, 2012) who in their field experiment send out letters to local councilors in UK. Contrary to the study by Bergan (2009) they did not find a significant effect on the decision done by the councilors, however, they observed that the politicians forwarded information-rich letters to another expert. The results suggest that the politicians often take clues from the indirect lobbyist. For many people the perception of politics and lobbying is, that money can buy anything and exchange of favors is often a thing there. Großer et al. (2013) address this issue in an experiment. They constructed a game where a firm has an opportunity to influence to politicians and buy then political favor. They observe that substantial part of the subjects than to use tactical quid pro quo agreements and that those agreements are also dependent on time. Minozzi and Woon (2013) investigated different issues that might be part of lobbying. They focused their experiment on lies in the communication and competition. They found that even though that lying aversion plays a role there, the competition in the political environment crowds it out. This is in line with many critics of politics. 7

8 2 Experiment Studying the transparency or lobbying in the lab is in general not novel as there exist wide literature on both topics. However, what is the gap, that needs to be filled, it is an investigation of effects of transparency on the behavior either of the lobbyist or lobbied politicians. Studies by Kartal and Tremewan (2016) or (Agranov and Tergiman, 2014) already open the discussion and contribute to investigating different elements of lobbying and its connection to transparency. Their studies show how the principles of lobbying might, on one hand, be disturbed by existing transparency and that secrecy and limited transparency creates the gray zone of political decisions. We also want to contribute to the research on transparent lobbying and design a novel experiment. The design we suggest is to answer a research question: Does the transparency have an effect on choice of lobbied policy alternatives over not lobbied alternatives by politicians?. This research question may contribute to the discussion whether the lobbying is only an exchange of money for favourable action in a legal bounderies, or wheter the action of politician asre driven by the well being of others as well. Therefore we bring this issue to the laboratory a design a game where two players interact and lobbying is an endogenous part of the game. 2.1 Design Both players - a politician (P ) and a voter (V ) - are in an environment that exists in two states (W ). The politician opens the game. He has three actions to do, choose a lobbied policy (C = Ā l ), or not lobbied policies Ā or B. When P makes his action, he knows probabilities of the states of the world (p(a) =< 0.4; 0.6 >, p(b) = 1 p(a)). He receives payoff α > 0 if C = W and also he gets a payment l > 0 when he chooses Āl. In the second stage of the game, the voter V learns information about lobbying, the state of the world W and policy C. If C = W, also V receives payment δ > 0. He then decides whether he punishes the politician (β < 0; β = l) or not. 8

9 Therefore the payoffs of both players are calculated as: π(p ) = α + δ + l (1) π(v ) = β (2) To illustrate the effects of transparency, we implement three treatments. First treatment, or a baseline, is a game with no transparency. In this game, the decision of the politician to accept or not the payment l is never revealed to the second player V. In mandatory transparency treatment, the information whether politician accepted the payment or not is always shown to the second player. The politician cannot hide the decision in any way. In the third treatment, voluntary transparency treatment, only the politician decides, whether the information that he accepted or not the payment l is shown to the voter or not. 2.2 Procedures The experiment has been by today conducted only as a pilot session. 22 subjects participated the session, half of them were female and all of them were students of different faculties of Masaryk University. The experiment was organized und administrated with the software hroot (Bock et al., 2014) and it was coded in z-tree (Fischbacher, 2007). Lasted about one hour and participants have played the game for 20 periods in partner matching setting. For the pilot, within subjects treatment testing was used. The decision-making screens are shown in Appendix A. The subjects were paid based on 3 randomly selected periods and earned in average Kč (about e6.5 ). To illustrate the game and settings used for the pilot study see the Figure 1. We used parameters l = {0, 20}, α = {0, 40}, β = { 20, 0} and δ = {0, 50}. 9

10 P C = Āl C = Ā C = B W = A W = B W = A W = B W = A W = B V β = l β = 0 ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) Figure 1: Game Tree 3 Preliminary Results The data presented in this paper draft are limited only to the results of the pilot session. The explanatory power is therefore limited and might change as we adjust the design to comments at different meetings. 3.1 Politician First, we look at the behavior of the politician. We assume that in the setup of our game, the politician will be less likely to accept the lobbied alternative in the mandatory transparency treatment than in no transparency treatment. We also assume, that the politician is a risk-averse player and will more likely choose the policy that is more likely in line with the probability of the world, i.e. if p(a) > 0.5 than C P = (ĀL Ā), and if p(b) > 0.5 than C P = ( B). We observed that in no transparency treatment the politician were less likely to accept the payment l (in 67% cases), on the other hand the politician accepted l in mandatory transparency treatment in 84% cases, as shows the Table 1. Using t-test we found the difference between treatments is statistically significant (t = , p-value ). These results do not meet our assumptions and we observe that the politician was also more likely to accept the lobbied 10

11 Treatment Obs Mean Std. Err. Std. Dev. [ 95% Conf. Interval] no transparency mandatory transparency combined diff Table 1: Means of the binary decision to accept payment from lobbyist by politician. Figure 2: Policy decisions of the politician. (0 - no transparency treatment, 1 - mandatory transparency treatment) alternative in mandatory transparency treatment (Figure 2). That might be explained by our second assumption, that the decisions are driven by probabilities of W. When we take a look at the distribution of percentual decision to accept the lobbied policy alternative Āl over the probability of the state of the world A we see a different behavior between treatments. In the mandatory transparency treatment, the politicians always chose the Āl if p(a) > 0.5, however it was not the case in the second treatment. To understand the behavior better, we focus on other variables as well. A probit regression of the decision to choose the lobbied policy alternative Āl (see Table 2) gives us a better picture. We see that the subjects were mostly driven by the probability of two states of the world and were more likely to choose Āl if the state of the world A was more probable to occur. What also played a role, was whether the politician was punished by the voter in a previous period (therefore the N in the table is smaller than 220 as period 1 was not included). 11

12 Figure 3: Means of the decision to choose lobbied alternative over the probability of world state A. (1) (2) (3) p(a) 0.150*** 0.152*** 0.154*** (0.03) (0.03) (0.03) Treatment (0.46) (0.47) (0.48) Punished (t-1) 0.639* 0.619* 0.676* (0.28) (0.29) (0.31) Period (0.02) (0.02) Male (0.46) const *** *** *** (1.51) (1.56) (1.65) R N Subjects Standard errors are reported in parentheses. *, **, *** indicates significance at the 90%, 95%, and 99% level Table 2: Probit regression of the decision to choose lobbied alternative. 12

13 3.2 Citizen The only action the citizens had was to either punish the politician or not. The decision does not cause any other costs for him, the consequences fall only on the politician. In out experimental design, we assume that the citizen punishes the politician more often in the transparency treatment, once he observes that the politician accepted the payment and the probability of W (A) was lower than 50%. We observe in average 26% of citizen to punish the politician in no transaprency treatment compared to 38.3% in other treatment (Table 3). The difference is statistically significant on 5% level (t = , p-value ). Treatment Obs Mean Std. Err. Std. Dev. [ 95% Conf. Interval] no transparency mandatory transparency combined diff Table 3: Means of the binary decision to punish the politician. The differences are also observable in the distribution of the proportion of citizen who punishes over the p(a) (Figure 4). However, if we take a look at the differences over periods, we do not observe any difference. Figure 4: Means of the decision to punish politician over the probability of world state A. 13

14 Figure 5: Means of the decision to punish politician over the period. (1) (2) (3) p(a) (0.03) (0.03) (0.03) Treatment (0.29) (0.29) (0.32) Punished (t-1) 0.839* 0.824* 0.815* (0.41) (0.41) (0.40) Period (0.02) (0.02) Male (0.33) const (1.41) (1.28) (1.35) R N Subjects Standard errors are reported in parentheses. *, **, *** indicates significance at the 90%, 95%, and 99% level Table 4: Probit regression of the decision to punish the politician. If we want to explain the behavior with other variables, we found statistically significant importance on their on the decision from the previous round. On the whole sample, those who punished in previous period were more likely to punish in the particular period (Table 4). To observe the effect of the decision of the politician to accept the lobbied payment, we can analyze only the data for mandatory transparency treatment (Figure 5). where we do not find the importance of such decision, however in this treatment the probability p(a) played a role. This finding might be caused by a limited number of observations and fact, that we conducted only one pilot session, therefore we do not tend to address it. 14

15 (1) (2) (3) Lobby (0.79) (0.79) (0.77) p(a) *** *** *** (0.03) (0.02) (0.02) Punished (t-1) 1.734** 1.744** 1.737** (0.64) (0.66) (0.65) Period (0.02) (0.02) Male (0.49) const 4.090*** 4.177*** 4.091*** (1.06) (0.79) (0.86) R N Subjects Standard errors are reported in parentheses. *, **, *** indicates significance at the 90%, 95%, and 99% level Table 5: Probit regression of the decision to punish the politician in transparent treatment. 4 Conclusion To make any conclusions based on the pilot study only would be audacious, however, we have shown that there exists a way to study the effects of transparency on decision making of a policy maker and his voter. We believe that we will contribute to the existing debate with our own experimental design. References Agranov, M. and C. Tergiman (2014). Transparency versus Back-Room Deals in Bargaining. Anctil, R. M., J. Dickhaut, C. Kanodia, and B. Shapiro (2004). Information transparency and coordination failure: Theory and experiment. Journal of Accounting Research 42(2),

16 Ball, R. (1991). Political lobbying as welfare improving signaling. University of California, Berkeley. Mimeograph. Bergan, D. E. (2009). Does Grassroots Lobbying Work?: A Field Experiment Measuring the Effects of an Lobbying Campaign on Legislative Behavior. American Politics Research 37, Bock, O., I. Baetge, and A. Nicklisch (2014). hroot: Hamburg registration and organization online tool. European Economic Review 71, Campos, N. F. and F. Giovannoni (2007). Lobbying, corruption and political influence. Public choice 131(1), Carey, J. M. (2013). Transparency & Legislative Behavior. In APSA 2011 Annual Meeting Paper. Côté, A. C. (2006). Lobbying and the Public Interest. Canadian Parliamentary Review. Fischbacher, U. (2007). Z-Tree: Zurich toolbox for ready-made economic experiments. Experimental Economics 10(2), Frechette, G., J. H. Kagel, and S. F. Lehrer (2003). Bargaining in Legislatures: An Experimental Investigation of Open versus Closed Amendment Rules. American Political Science Review 97(2), Fréchette, G. R. and A. Schotter (2015). Handbook of Experimental Economic Methodology, Volume 30. Grimmelikhuijsen, S. (2012). Linking transparency, knowledge and citizen trust in government: an experiment. International Review of Administrative Sciences 78(1), Großer, J., E. Reuben, and A. Tymula (2013). Political quid pro quo agreements: An experimental study. American Journal of Political Science 57(3), Grossman, G. M. and E. Helpman (2001). Special Interest Politics. Gene,

17 Helland, L. (2008). Lobbying with conflicting interests: Norwegian local-central relations. European Journal of Political Research 47(2), Kartal, M. and J. Tremewan (2016). An offer you can refuse: the effect of transparency with endogenous conflict of interest. Working paper. Kretschmer, H. and H.-J. Schmedes (2010). Enhancing transparency in EU lobbying? Krueger, A. O. (1974). The Political Economy of Rent-Seeking Society. The American Economic Review 64(3), Lambsdorff, J. G. (2002). Corruption and rent-seeking. Public choice 113(1), Lohmann, S. (1995). Information, access, and contributions: A signaling model of lobbying. Public Choice 85(3-4), Malesky, E., P. Schuler, and A. Tran (2012). The adverse effects of sunshine: a field experiment on legislative transparency in an authoritarian assembly. American Political Science Review 106(04), Minozzi, W. and J. Woon (2013). Lying aversion, lobbying, and context in a strategic communication experiment. Journal of Theoretical Politics 25(3), Mueller, D. C. (2003). Public Choice III, Volume 30. Peisakhin, L. and P. Pinto (2010). Is transparency an effective anti-corruption strategy? Evidence from a field experiment in India. Regulation and Governance 4(3), Potters, J. and F. van Winden (1996). Comparative statics of a signaling game: An experimental study. International Journal of Game Theory 25(3), Potters, J. and F. van Winden (2000). Professionals and students in a lobbying experiment - Professional rules of conduct and subject surrogacy. Journal of Economic Behavior & Organization 43(4),

18 Richardson, L. and P. John (2012). Who Listens to the Grass Roots? A Field Experiment on Informational Lobbying in the UK. British Journal of Politics and International Relations 14(4), Snyder, J. M. (1991). ON BUYING LEGISLATURES. Economics & Politics 3(2), Tullock, G. (1967). THE WELFARE COSTS OF TARIFFS, MONOPOLIES, AND THEFT. Economic Inquiry 5(3), Tullock, G. (1980). Rent Seeking as a Negative-Sum Game. Toward a Theory of the Rent Seeking Society, edited by Buchanan JM, Tollison RD, Tullock G.. College Station: Texas A&M University. 18

19 Appendix A Screens A.1 Politican Figure 6: Decison making screen - politician I Figure 7: Decison making screen - politician II 19

20 Figure 8: Decison making screen - politician III A.2 Citizen Figure 9: Decison making screen - citizen 20