The political economy of agricultural liberalization in Central and Eastern Europe: An empirical analysis

Food Policy 49 (2014) 332–346

Contents lists available at ScienceDirect

Food Policy journal homepage: www.elsevier.com/locate/foodpol

The political economy of agricultural liberalization in Central and Eastern Europe: An empirical analysis Jeroen Klomp ⇑ Wageningen University, The Netherlands

a r t i c l e

i n f o

Article history: Received 16 January 2013 Received in revised form 8 April 2014 Accepted 22 August 2014

Keywords: Agricultural liberalization Election cycles Government ideology Central and Eastern Europe

a b s t r a c t We examine the effect of upcoming elections and government ideology on agricultural liberalization in Central and Eastern Europe countries in the post-communist period. Our results suggest first that prices and markets liberalization and land market privatization are manipulated in pre-election periods to secure re-election by favouring farmers. Second, we find no evidence that reforms in the agro-process industry, rural finance or institutional environment are affected by upcoming elections. Third, we demonstrate that right-wing governments protect the interest of the agricultural sector more than left-wing governments by affecting the speed of price and market reforms, privatization in the agro-processing industry and land market privatization. Finally, we demonstrate that liberalization the agricultural sector is partly retarded by nationalistic governments. Ó 2014 Elsevier Ltd. All rights reserved.

Introduction Since late 1989, when communism collapsed, most Central and Eastern European countries (CEEC’s) have gone through some considerable democratic reforms. One popular view is that the establishment of democratic institutions has stimulated economic liberalization in the post-communist world, especially in the agricultural sector (cf. Johnson, 1995; Kennedy, 1999; Rozelle and Swinnen, 2004; Swinnen and Rozelle, 2009; Swinnen, 2002; Deininger, 2003). Agriculture in Central and Eastern European countries is a much more important component of the economy than in industrialized countries. It traditionally accounted for 15– 20% of GDP and total employment, compared to only 2–3% in the EU. In the pre-reform period, agriculture was heavily subsidized in a number of CEE countries, such as the Czech Republic and Hungary, to stabilize farmers’ income, ensuring food security and protecting farmers from import competition. Conversely, in a number of other CEE countries, the agriculture sector was heavily dominated by state ownership and taxation of the production (Cungu and Swinnen, 1999; Macours and Swinnen, 2002). In the beginning of the 90s, most CEEC’s began to open up their agricultural markets and start declining government involvement. From that point onward, trade was more liberalized, price distortions were

⇑ Address: Wageningen University, Development Economics Group, P.O. Box 8130, 6700 EW, Wageningen, The Netherlands. http://dx.doi.org/10.1016/j.foodpol.2014.08.003 0306-9192/Ó 2014 Elsevier Ltd. All rights reserved.

removed, property rights were privatized and the level of support provided to agriculture declined drastically.1 Meanwhile, agricultural reforms are not always taken to spur development only but also for political purposes (cf. Park and Jensen, 2007; Klomp and De Haan, 2013). Incumbents have powerful incentives to affect voters’ behavior by using socio-economic reforms to secure the support of particular constituencies. This is especially the case when elections are at hand and voters base their behavior on the recent past (Nordhaus, 1975). According to Anderson and Hayami (1986), there are two main arguments why the incumbent government may protect or support the agricultural sector for re-election purposes at the expense of other voter groups. First, in view of the size of the agricultural sector, farmers have an organizational advantage over other diffused interests and may therefore be more successful in mobilizing campaign contributions and votes. Second, when income support is given to lower prices for agricultural products, this increases taxpayers’ real income and thereby making the total tax burden associated with agricultural protection socially affordable. Thus, if a general election is approaching, the government may be reluctant to implement structural liberalization reforms on farmers that have very high short-term costs (Pitlik and Wirth, 2003). This view is empirically supported by Thies and Porche (2007) who find a

1 According to Swinnen and Rozelle (2009) and Bjørnskov and Potrafke (2011) some countries chose shock therapy, others succeeded in implementing more gradual reforms, while a few either took no reform measures or rolled reforms back swiftly after beginning to implement them.

J. Klomp / Food Policy 49 (2014) 332–346

positive and significant effect of upcoming elections on the degree of protection of the agricultural sector. Likewise, Klomp and De Haan (2013) and Park and Jensen (2007) demonstrate that agricultural income support increases in (pre-)election years. In addition, agricultural reforms may not only be affected by upcoming elections but also by the government’s ideology. According to the traditional partisan theory, left and right-wing political parties have different preferences as to the size and scope of government influence (Potrafke, 2011). Likewise, they may have different views on agricultural policy. Empirical evidence suggests that right-wing governments follow more protectionist policies and provide more support to agricultural producers than left-wing governments due to constituency motives (Olper, 2007; Swinnen, 2010a). Up so far most studies explore only the political economy of agricultural protection or land market reforms (cf. Thies and Porche, 2007; Klomp and De Haan, 2013; Giuliano et al., 2013; Swinnen, 1999), thereby neglecting reforms in other policy areas. In our study we argue instead that agricultural liberalization consists out of multiple policies related to (1) price and market liberalization; (2) land market privatization; (3) agro-processing and input supply privatization; (4) rural finance liberalization and (5) market institutions liberalization. Our contribution to the empirical literature is by examining whether election cycles and the government ideology affect the multiple dimensions of agricultural liberalization in CEE countries. It is well possible that these cycles have diverse or even opposite effects on the various dimensions of liberalization since some reforms improve the welfare of farmers by reducing production costs through enhanced efficiency, while others may harm farmers through a loss of trade protection or financial support. To explore these impacts, we apply a dynamic panel model including about 20 Central and Eastern European countries over the period 1996–2005. We use the different dimensions of liberalization reported by the World Bank which refer to the five dimensions outlined above. We address the potential endogeneity problems by presenting a system-GMM model. After testing for the sensitivity of the results, we can draw a number of conclusions. First, liberalization in prices and markets is slowed down or partly reversed under the influence of upcoming elections in countries which subsidize the agricultural sector or have open markets. However, in countries where the agricultural sector previously was taxed or under state control, there we find a positive election impact on prices and markets reforms. This latter result can be explained that in these countries, farmers benefit from a more open trade system with less distortions. Second, governments stimulate land market reforms during an election year by allowing for more private ownership. Third, we do not find any evidence of electoral cycles in reforms in the agro-processing industry, rural finance and institutions as reforming these dimensions at a short notice is politically difficult or the individual benefits for farmers may be too small to signal competence of the cabinet. Consequently, reforms affecting these dimensions are less appropriate as a re-election instrument. Fourth, right-wing governments support or protect the interests of the agricultural sector more than left-wing governments. To be more precise, Former Soviet Union member states ruled by a right-wing cabinet open up their markets more often and decline the agricultural taxation, while in non-former Soviet Union member states right-wing governments start subsidizing farmers more. Fifth, nationalist-led governments are associated with slower transition speeds to re-enforce state control. Finally, it turns out that the effect of election cycles in price and market liberalization is partly conditional on the political system in place. In particular, the election effect is stronger under a mixed electoral system compared to a proportional system. Our interpretation of this finding is that a mixed system gives the incumbent a stronger incentive to target transfers to particular interest groups, like agricultural producers than under a strict proportional system.

333

The remainder of the paper is structured as follows. Section ‘Theoretical background’ reviews the theoretical background underlying our hypotheses. Section ‘Data and methodology’ describes the data and methodology used, while Section ‘Main results’ shows our results for the influence of election cycles and government ideology on reforms in the agricultural sector. Section ‘Robustness analysis’ presents a sensitivity analysis, while the final section offers the conclusions. Theoretical background Intuitively one could argue that if a country is democratic and rural dwellers constitute a large segment of the voting population, politicians have powerful incentives to cater to the interests of farmers. Conversely, when large shares of the population are poor and live in urban areas, they demand that the government protects their interests by adopting policies that lower the costs of food. However, according to the current empirical evidence these pictures do not represent reality. That is, in low-income countries the agricultural sector is often being taxed to the benefit of import competing sectors, the so-called development paradox (Bates and Block, 2011a,b; Swinnen, 2010b; Anderson et al., 2013). Up so far most of the literature on the political economy of the agricultural sector is only focused on explaining differences in agricultural protection measured by the degree of price support (cf. Olper, 2007; Olper and Raimondi, 2004; Olper et al., 2010; Klomp and De Haan, 2013; Thies and Porche, 2007) or ownership in the rural land market (cf. Giuliano et al., 2013; Swinnen, 1999).2 In our study instead we combine the electoral and partisan cycle theories and apply these theories to the various dimensions of liberalization reforms in the agricultural sector. The influence of election cycles on agricultural liberalization is theoretical not directly clear. On the one hand, politicians are more likely to conduct liberalization reforms in the wake of elections when the majority benefits in economic terms. For instance, reducing agricultural price support may tend to impose short-term costs on farmers, while generate economic benefits for society at large. On the other hand, politicians who undertake economic reforms as elections approach, risk alienating constituents who will bear the near-term costs of this reform. Besides, politicians may fear the risk not leaving enough time for the majority of voters to realize the corresponding gains (cf. Haggard and Webb, 1993; Frye and Mansfield, 2004). This problem becomes more severe if the short-term costs of liberalization are disproportionately borne by protectionist interest groups that are better organized, better informed, and more politically powerful than society at large, which is harmed by protection (Olson, 1965). The existing literature clearly indicates, that although, agricultural support in many countries benefits only a small subset of population, politicians may seek electoral support in the agricultural sector (cf. Thies and Porche, 2007; Klomp and De Haan, 2013). For instance, Thies and Porche (2007) find a positive and significant effect of upcoming elections on protection of the agricultural sector in OECD countries. One major limitation of the existing empirical studies is that it is only focused on the impact of political cycles on price support. However, according to Csaki (2000) the reform agenda in agriculture should address problems in at least five areas: (1) the removal of direct government intervention from agricultural markets and the creation of a market-compatible policy framework for the agrarian economy, including liberalization of prices and markets for farm products and inputs; (2) privatization of land and creation of new farming structures based on private ownership of land and productive assets; (3) creation of a competitive environment for 2 For instance, Giuliano et al. (2013) conclude that democracy has a significant positive impact on the implementation of agricultural reforms in private ownership and market prices.

334

J. Klomp / Food Policy 49 (2014) 332–346

agriculture, including privatization and demonopolization of agroprocessing, input supply, services, and trade; (4) creation of a rural financial system serving the needs of privatized agriculture and agricultural services and (5) institutional reform, creating institutions and public services that support a functioning market economy.3 One can argue that the first two dimensions are more related to reforms that directly affect production decisions made by farmers, while the latter dimensions are associated with the market environment in which farmers operate. However, it is theoretically not direct clear in what direction agricultural liberalization moves since political cycles may have different, even opposite impacts on the five elements of the reform agenda. To be more precise, the exact impact is determined by three factors. First, the reforms considered vary in the way they affect the economic wellbeing of farmers. While some reforms harm farmers due to the loss of protection or support, whereas others may create benefits through reduced production costs. Second, the intensity of reforms may differ since some reforms deliver greater gains or losses to farmers than others. Third, some dimensions are more difficult to reform at a short notice than others, for instance due to political constraints, making them less appropriate as a re-election instrument. To illustrate the theoretically impact of political cycles on agricultural liberalization, we discuss each dimension separately and derive the corresponding hypothesis that is later on tested in the empirical section. The first dimension we discuss is related to liberalization in prices and markets. The empirical evidence so far demonstrates that politicians behave rather opportunistically before elections to increase their re-election prospects by giving income support to farmers (cf. Klomp and De Haan, 2013; Park and Jensen, 2007) or increasing trade protection (Thies and Porche, 2007). When income support is given to lower prices for agricultural products, this increases taxpayers’ real income and thereby making the total tax burden associated with agricultural protection socially affordable. However, the ultimate outcome in CEEC’s might depend on the pre-reform situation (Macours and Swinnen (2000, 2002). Farmers may benefit from more liberalized prices markets when they are taxed, while they are being hurt when they were subsidized before. Based on this observation, we derive the following two hypotheses H1a. Prices and markets liberalization is slowed down or partly reversed when elections are upcoming in countries that subsidize the agricultural sector. H1b. Prices and markets liberalization is spurred when elections are upcoming in countries that tax the agricultural sector. The second dimension identified by Csaki (2000) refers to land market privatization. Swinnen (1999) argues that the land market reforms in the CEEC’s that started in the beginning of the 1990s are based on legal demands of the pre-collectivization landowners and social equity concerns. However, there is also a possibility that a part of these reforms are more politically motivated. The big-bang approach used to residue land right in a number of CEEC’s was an attempt to break the rural power and organizational structure of the communistic regime (Swinnen and Rozelle, 2009). According to the so-called capture theory politicians adopt policies that favor the industry or interest group over whose activities they have jurisdiction. For instance, Swinnen (1999) argues that many CEEC’s did not restitute land to former landowners who are no longer live in the country as these former owners do not longer belong to the

3 See for an extensive survey on the reforms on prices and markets, property rights and market institutions in Central and Eastern Europe Rozelle and Swinnen (2004).

domestic political constituency and there are no political gains in restituting the land to them. Electoral competition may promote efficiency-enhancing policies for certain targeted industries. It is well established in the literature that an increase in the security of property rights enhances growth (cf. Knack and Keefer, 1995; Deininger, 2003). For instance, Rozelle and Swinnen (2004) note that more than any other reform policy, property rights reform are responsible for the rapid growth in agricultural productivity in many CEEC’s. Thus, one can than argue that politicians are more inclined to implement a land market reform in a pre-election year as it potentially stimulates agricultural production and signals therefore the competence of the incumbent cabinet. H2. Land market privatization is increased when elections are upcoming. The third dimension is associated with reforms in the agro-processing industry. The theoretical impact of elections on liberalization reforms in the agro-processing industry is ambiguous. On one hand, when due to privatization and demonopolization of the agroprocessing industry this market becomes more open for competition, especially from foreign agro-processing enterprises, it may enhance efficiency in this sector. For instance, according to Kiewiet (2000) state-owned enterprises are rarely known for their efficiency of their operations or the quality of their products. Farmers may benefit from lower processing costs or increased demand from foreign enterprises.4 On the other hand, the domestic agroprocessing industry itself may lobby for more protection by increasing entry barriers or production subsidies. Besides state-owned enterprises often operate as public employment programs and objects of political patronage in many CEEC’s. Layoffs from stateowned enterprises are thus unpalatable (Kiewiet, 2000). Based on these observations, we can argue that H3. Agro-processing industry privatization is either increased or decreased when elections are upcoming. The fourth dimension of the reform agenda is related to the openness of the agricultural financial markets. Since 1990s, the rural banking sector have gone through some major reforms leading to a rural banking network where micro-banking institutions are one of the most important financial intermediaries, while the traditional commercial banks play a subordinate role. Nevertheless, a lack of agri-rural financing is still one of the most serious constraints to productivity growth in the agricultural sector (Buchenrieder et al., 2009; Swinnen and Gow, 1999). The impact of upcoming elections on liberalizing rural financial markets is not directly clear as it consist out of multiple policies which may have opposite effects. For instance, a more open financial system would increases competition among intermediaries which may depress interest rates on loans when there is a level playing field. However, a more liberalized financial market would conversely also imply that low-cost credit provided by the government through a subsidy on the principal and interest rate provided in many CEEC’s should be abolished (Ciaian and Swinnen, 2007). Besides, Buchenrieder et al. (2009) argue that there is a special class of borrowers, notably state-owned farms, large-scale borrowers, and clients with strong lobbying powers and close ties to the banking system that benefit from additional favors supplied by banks and lobby therefore for less rigorous liberalization. We argue that H4. Rural banking industry liberalization is either increased or decreased when elections are upcoming.

4 Not only farmers, but the society as a whole may benefit from lower production costs.

J. Klomp / Food Policy 49 (2014) 332–346

Finally, in order to function efficiently agricultural markets require supporting institutions to ensure competition, define and enforce property rights and contracts, ensure access to credit and finance, and provide information (Rozelle and Swinnen, 2004). The impact of upcoming elections on changes in the institutional agricultural environment is presumable small as reforming institutions is often only feasible in the long run due to the political constraints present. For instance, agricultural policies and institutions are often subject to some kind of ratchet effect which makes them sticky and even harder to reform (Thies and Porche, 2007). According to Swinnen and Rozelle (2009) most successful transitions have not gone straight from planning to decentralized market-based exchange but move gradually. Therefore it less likely that institutional reforms are manipulated as a re-election strategy and are therefore less subject to an electoral cycle. H5. The agricultural institutional environment is not affected by upcoming elections. The second strand of literature on which we build is the partisan theory that focuses on different preferences on market-orientated reforms of left and right-wing parties. These differences are in line with the interests of the constituencies of the political parties (Hibbs, 1992). Theoretically, the impact of ideology on agricultural protection is not directly clear. On the one hand, left-wing governments favor more state intervention, more income redistribution and expansionary public policies. In contrast, right-wing governments believe in the free market and favor less state intervention, so more privatization and liberalization. The empirical studies by Bjørnskov and Potrafke (2011) Bortolotti and Pinotti (2008) and Bortolotti et al. (2001, 2003) show clearly that for many countries the privatization process was forced by right-wing and market-oriented governments. A rational explanation for implementing market-orientated reforms may be the forward-looking behavior of politicians aiming at gaining future support from the constituencies of owners of newly privatized companies. However, on the other hand, when we presume—in line with the traditional partisan theory—that a left-wing government represents the interests of workers than it will reduce agricultural protection as workers spend a large share of their income on food. However, in case of CEEC’s the large-scale farms employ a large number of workers which all benefit from more protection. This makes it possible that in these particular countries left-wing governments are more inclined to support the agricultural sector. Moreover, the policy stance of a right-wing government depends crucially on which constituency the right-wing government represents. If it represents the interests of capitalists, it will reduce agricultural protection as lower agricultural prices have less impact on the wages earned by the workers. However, when right-wing governments give priority to the interests of land owners, than they will increase agricultural protection as landowners will benefit from more protection (Swinnen, 2010a,b). H6. Agricultural liberalization increases (decreases) more with a right-wing cabinet than with a left-wing cabinet. Thus, the question whether right-wing governments are more or less protective may ultimately be an empirical one. There is substantive evidence for industrial countries that farmers vote for right-wing parties (see Lewis-Beck, 1977 and references cited therein). The evidence of Olper (2007), Klomp and De Haan (2013) and Swinnen (2010a) report that right-wing governments are more protectionist and spend more on support to the agricultural sector than left-wing governments. In contrast, Bates (1989) finds that left-wing governments support the agricultural sector in more unequal societies.

335

In addition, the cabinet in several Central and Eastern European political systems is made up by at least one strongly nationalist party. Many of these nationalistic parties have still often ties with the former communistic parties. One can argue that nationalistic governments, mainly due to their strong preferences against globalization, provide more protection to agricultural sector, regardless if there is a left or right-wing government in office. As these kind of governments are more related with collectiveness, this may have in particular implications for privatization in the land market and the agro-process industry. For instance, Bjørnskov and Potrafke (2011) demonstrate that privatization in CEE countries was retarded by nationalistic governments. Likewise, Fidrmuc (2000) shows for a number of CEEC’s that there is a positive relationship between the agricultural labor share and the vote share received by nationalist parties and a negative relationship with the votes casted to pro-reform political parties. H7. Agricultural liberalization is reduced in countries ruled by a nationalistic cabinet.

Data and methodology Data As our aim is to consider the impact of political cycles on liberalization reforms in the agricultural sector, we have to quantify these reforms. We used indices on agricultural sector liberalization and privatization reported by the World Bank which incorporates the dimensions of liberalization outlined by Csaki (2000): (1) price and market liberalization; (2) land market privatization; (3) agroprocessing and input supply privatization; (4) rural finance liberalization and (5) liberalization in market institutions. The indices are measured from 1 (centrally planned economy) to 10 (completed market reforms). Table 1 reports the average level of agricultural liberalization and the annual percentage change herein, while Fig. 1 shows the development of the different dimensions throughout our period of analysis. There are large differences in the degree of liberalization across countries. While in the Czech Republic and Hungary the agricultural sector is highly liberalized, whereas in Uzbekistan and Turkmenistan the agricultural sector is still under state control. One rational explanation is that the previous countries reformed much faster due to their application or prospect to become an EU member. Moreover, also the degree of liberalization between the different dimensions differs significantly. Prices and markets are much more liberalized compared to the rural financial system or the institutional framework.5 To proxy the political budget cycle, we use an election variable suggested by Franzese (2000) that takes the timing of an election in the course of a year into account. Compared to using a dummy that is one in election years and zero otherwise, which is common in this type of research, our proxy reduces measurement error.6 It is calculated as M/12 in an election year and (12  M)/12 in a preelection year, where M is the month of the election. In all other years its value is set to zero. The election data is taken from electionsource.org and various issues of the Political Handbook of the World. 5 One concern with the data used is that the observations of a particular country stops when it enters the EU. This explains why there is a drop in the graph in 2004 as in that year the Czech Republic, Hungary, Slovakia and Slovenia became a member of the EU. Because of this, we are not able to capture the impact of tariff protection and the CAP subsidy when being a EU member. However, these latter policy measures may be less subject to political cycles of individual EU members as they are taken centralized by the European Commission. 6 An election in January gives a value of 1 for the election dummy. However, to be effective public policy arguably had to be adjusted earlier so that the election dummy may not properly identify the election effect.

336

J. Klomp / Food Policy 49 (2014) 332–346

Table 1 Descriptive statistics agricultural liberalization. % annual change

Albania Armenia Azerbaijan Belarus Bosnia Bulgaria Croatia Czech Estonia Georgia Hungary Kazakhstan Kyrgyz Latvia Lithuania Macedonia Moldova Poland Romania Russia Serbia Slovak Slovenia Tajikistan Turkmenistan Ukraine Uzbekistan Average Standard deviation

Average level of liberalization 1996–2005

Prices and markets

Land market

Agroprocessing

Rural finance

Institutions

Prices and markets

Land market

Agroprocessing

Rural finance

Institutions

0.17 1.79 0.74 2.08 4.42 5.90 5.65 0.23 1.67 1.96 0.23 1.01 3.87 4.46 5.06 1.79 1.56 0.53 3.87 1.79 26.19 4.46 2.08 3.54 2.08 1.07 4.79

1.56 0.57 1.98 12.50 2.72 1.96 6.37 2.08 9.49 1.19 1.85 8.61 3.87 1.85 2.08 0.00 2.31 2.31 1.79 2.50 10.86 4.46 1.85 19.46 2.08 0.42 28.13

0.00 3.35 8.75 10.42 2.38 8.35 3.87 3.94 6.32 2.50 1.85 1.49 2.08 6.32 4.46 1.26 1.79 6.32 5.43 3.35 23.52 3.94 4.17 1.88 12.50 2.08 32.29

12.92 0.00 5.63 0.00 2.38 9.49 2.08 2.08 4.46 0.30 2.08 0.71 2.38 4.46 5.16 7.71 4.58 5.16 4.29 2.92 17.33 2.08 2.38 6.88 0.00 5.10 22.92

4.58 2.38 3.13 22.92 0.00 7.93 4.39 3.94 1.85 7.71 2.08 0.00 3.21 2.08 4.46 2.08 3.75 0.00 9.49 0.63 15.86 4.46 2.31 2.50 4.17 7.29 5.21

8.11 7.67 7.11 2.33 6.75 8.33 7.11 8.86 9.14 8.00 8.71 6.22 7.22 8.57 7.71 7.89 6.78 8.14 7.33 6.11 6.50 7.86 8.86 5.44 2.11 6.11 3.89

8.22 8.00 7.22 1.78 6.38 8.00 7.00 8.57 8.14 6.44 9.14 5.78 7.22 9.14 8.57 7.00 7.00 8.43 7.78 5.22 6.50 8.00 9.14 5.33 2.56 5.56 3.56

8.00 7.89 6.44 2.33 6.13 8.00 7.33 9.43 8.71 4.89 9.86 6.44 6.22 8.71 8.14 6.67 6.22 8.86 7.67 7.67 5.83 8.86 9.14 4.78 1.56 7.00 3.11

5.78 7.00 5.44 2.00 6.25 6.44 6.22 8.86 8.43 6.11 8.86 5.89 6.56 8.43 6.71 5.33 6.11 7.00 6.67 5.56 4.83 8.29 8.00 3.11 1.00 6.00 2.22

6.22 7.11 4.89 1.67 5.00 7.67 7.67 8.86 9.29 4.56 8.57 5.00 5.78 8.57 8.29 6.89 4.44 8.00 6.33 4.89 6.33 7.57 9.00 4.11 2.33 4.11 3.33

2.99 5.08

4.91 6.37

5.44 7.45

5.02 5.25

3.69 5.72

7.00 1.78

6.88 1.89

6.88 2.07

6.04 1.99

6.17 2.07

This table shows the annual change and the average level of the five dimensions of agricultural liberalization between 1996 and 2005.

We only include elections if the government has sufficient time to change its public policies. When there are, for instance, elections shortly after the fall of a cabinet, the government may have little opportunity to change their policy. An election is therefore only included if the election is held on the fixed date (year) specified by the constitution, or if the election occurs in the last year of a constitutionally fixed term for the legislature. Also when an election is announced more than one year in advance, it is taken up in the analysis (Shi and Svensson, 2006). To proxy partisan cycles we use an ideology index proposed by Potrafke (2011). This index places the cabinet on a left-right scale with values between 1 and 5. It takes the value 1 (5) if the share of governing right-wing (left-wing) parties in terms of seats in the cabinet and in parliament is larger than 2/3, and 2 (4) if it is between 1/3 and 2/3. The index is 3 if the share of center parties is 50%, or if the left and right-wing parties form a coalition government that is not dominated by one side or the other. We base our partisan measure on the data provided by the Database of Political Institutions (Beck et al., 2001).7 In addition, as already mentioned above, a number of Central and Eastern European countries have at least one strongly nationalist party as a member of the coalition. To capture this notion, we include a dummy variable if the government in a particular country-year is recognized as a nationalistic government by the Database of Political Institutions. A government is identified as nationalistic when a primary component of the executive party platform is the creation or defense of a national or ethnic identity.

7 Years in which a new government took over are labeled based on the ideological position of the government that was in office for most of the year concerned. For instance, when a right-wing government replaces a left-wing government in August, this year is labeled as left-wing.

Model Following Shi and Svensson (2006), we estimate the relationship between political cycles and agricultural liberalization using the following dynamic panel model based on an unbalanced dataset between 1996 and 2005 for about 20 CEE countries. We only consider country-years with a minimum level of democracy as the political cycle theories presumes that competitive elections take place. We therefore only include country-years with a Polity IV democracy score of at least five. Table A1 in the appendix shows all countries included.

ln agrilibit ¼ ait þ c ln agrilibit1 þ

J X bk xkit1 þ l cycleit þ eit

ð1Þ

j¼1

where agrilib is the degree of liberalization of the agricultural sector represented by the five indices outlined above (taken in logarithms) in country i in year t, xkit is a vector of k control variables, cycleit refers to the political cycle indicator outlined above (election or ideology variable), and eit is an error term. The parameter ait is a country-specific time trend placing the emphasis for identification of effects on the within country variation over time, for instance, to capture partly the idea that many countries have already prepared the entry to the EU a long time ago by implementing the necessary reforms.8 An alternative way of writing the previous equation is 8 One may argue that the impact of pre-EU access policies may differ substantially among the various reforms, as some reforms, like prices and markets liberalization, are a more important entry requirement of the EU than others (see also Csaki, 2000). This view is also supported by our empirical evidence as the magnitude of the country-specific time trend significantly differs between the different types of reforms considered.

337

J. Klomp / Food Policy 49 (2014) 332–346

and Blundell and Bond (1998) propose to combine the above differenced regression with the original regression in levels. The instruments for the regression in differences are those described above, while the instruments for the regression in levels are the lagged differences of the dependent variables. Formally, the additional moment condition is the following.

E½D2 ln agrilibits ; ðait þ eit Þ ¼ 0 s P 1 E½Dxkits ; ðait þ eit Þ ¼ 0

Fig. 1. Agricultural liberalization in 27 CEE countries, 1996–2005. Average agricultural liberalization weighted by the share of the agricultural sector in GDP. A score of 1 indicates centrally planned economy, while a score of 10 demonstrates completed market reforms.

ln agrilibit  ln agrilibit1 ¼ ait þ ðc  1Þ ln agrilibit1 þ

J X bk xkit1 þ l cycleit þ eit

ð2Þ

j¼1

In this specification, the dependent variable is the log-change of the degree of liberalization in the agricultural sector. The presence of lagged dependent variables and the country-specific effects in Eq. (2) render the OLS estimator to be biased. The magnitude of the bias depends on the length of the time series and only when it goes to infinity will the fixed effects estimator be consistent. Since the average number of observations across countries in our sample is rather small, the bias of the fixed effects estimator may be non-negligible. As a solution, we can eliminate the country-specific effects by taking first-differences

Dðlnagrilibit  lnagrilibit1 Þ ¼ ðait  ait1 Þ þ ðc  1ÞD lnagrilibit1 þ

J X

bk Dxkit1 þ l Dcycleit þ Deit

ð3Þ

j¼1

Estimation of the first-difference equation requires an instrumental variable procedure to correct for endogeneity and the correlation between the lagged difference of the dependent variable and the lagged error term. Arellano and Bond (1991) note that under the assumption that the error term is not serially correlated, lagged values of the dependent variable are valid instruments for the transformed lagged dependent variables. As a result, we can use the second and higher-order lags of the endogenous and dependent variable and the first-difference of the exogenous variables as instruments in the estimation of the first-difference equation if the error term is serially uncorrelated. For the control variables, we assume that they are weakly exogenous. That is, they are uncorrelated with future realizations of the error term. Thus, the GMM dynamic first-difference estimator uses the following linear moment conditions,

E½D ln agrilibits ; Deit  ¼ 0 s P 2; t ¼ 3:::T E½xkits ; Deit  ¼ 0

s P 2; t ¼ 3:::T

ð4Þ

where s are the number of lags chosen. We use Dcycleit as its own instrument. The moment conditions above are sufficient to estimate the parameters of the model. When the explanatory variables are persistent over time, lagged levels of these variables are weak instruments for the regression equation in differences. In order to increase the precision of the estimates, Arellano and Bover (1995)

sP1

ð5Þ

Combining the moment conditions for the difference and level equations yields the system-GMM estimator. As long as the model is over-identified, the validity of the assumptions underlying both the difference and the system estimators can be tested through Sargan tests of orthogonally between the instruments and the residuals and through tests of second- or higher order residual autocorrelation.9 Besides, we consider the Hansen test of over-identifying restrictions, where the null hypothesis is that the instruments are uncorrelated with the residuals. Although the system-GMM estimator is asymptotically more efficient, the standard error estimates from the two-step covariance estimation tend to be severely downward biased (Arellano and Bond, 1991; Blundell and Bond, 1998). We correct the bias using the finite sample correction of the two-step covariance matrix derived by Windmeijer (2005). The vector of control variables is largely based on previous studies on the political economy of the agricultural sector (cf. Bates and Block, 2011a,b; Klomp and De Haan, 2013; Olper, 2007). These variables are required to capture the role of policies and institutions, and helps to avoid an omitted variable bias. Table A2 in the Appendix offers a description and provides the sources used. First, we include a group of variables related to the economic performance of a particular country. We start by adding real GDP per capita (taken in logarithms) and the growth rate of real GDP per capita. These variables control for the level of development of a country and the change herein as this may influence voters’ preferences for liberalization and privatization (see Swinnen, 1994; Anderson, 1995). Inflation measured by the change of the GDP deflator is taken up as a measure of mismanagement in macroeconomic policies. Bad macroeconomic policies may increase the opposition to ongoing liberalization programs. In addition, we include trade openness measured as import plus export as a share of GDP. One of the most powerful arguments for a halt on liberalization is the welfare loss due to trade liberalization when an economy or sector is not competitive. In addition, we account for pressure by the International Monetary Fund (IMF) for economic reforms. We include a dummy variable, indicating whether country has a structural adjustment loan from the IMF in a particular year. The second group of variables we consider is related to the agricultural dependence of the economy of a particular country. We include the rural population share. On the one hand, it is probably more difficult for the government to move to a more market-oriented agricultural sector in a country with a larger rural population than in a country with a large urban population (Henning et al., 2011). On the other hand, small groups are more efficient in lobbying and collective action (Olson, 1965).10 In addition, we include a number of measures to control for the economic importance of the agricultural sector in a particular country: land per capita (measured

9 Specifically, we used the xtabond2 command implemented by Roodman (2006) in Stata. 10 In contrast, a case study done by Cungu and Swinnen (1999) indicates that while Albania has a large rural population share, the agricultural liberalization was rather fast.

338

J. Klomp / Food Policy 49 (2014) 332–346

by arable land in square kilometers divided by the total labor force working in the agricultural sector) and agricultural capital per capita (agricultural physical capital divided by the number of workers in the agricultural sector).11 We also include the share of the agricultural sector in GDP and the net agricultural export (as a share to total export) taken from the FAO as measures of the comparative advantage of this sector in a particular country.12 To capture reforms necessary to become a member of the EU, we include a count variable measuring the number of years before EU access. To overcome the problem of losing observations by including countries for which it is still unknown when they enter the EU, the particular countries receive the value 12, which is the maximum number of years in our dataset plus one. Finally, we include a set of political variables to take the general setting in which the cabinet is embedded into account. Following Swinnen (2010a) and Olper (2007), we add a variable to control for the degree of democracy by using the Polity IV scores. For instance, Falkowski and Olper (2012) report a positive correlation between the level of democracy and the degree of agricultural protection. Moreover, Olper et al. (2014) document that agricultural protection increases with the transition to a democracy13. According to Olper (2001) this positive correlation not per se caused by the level of democracy, but by the quality of the institutions that protect and enforce property rights. Persson and Tabellini (2002) argue that elections may have a different effect on public policy under proportional and majoritarian electoral rules. Proportional elections induce politicians to seek support from larger groups in the electorate via spending of which more than the majority benefits. It is then plausible to expect larger expansions under proportional electoral rules than under majority electoral rules. Though, Persson and Tabellini (2002) also argue that distributional policies targeting only a small sector of voters are more likely in majoritarian systems than in proportional systems because marginal districts are more important in majoritarian elections. Similar, Park and Jensen (2007) find that countries with electoral systems in which politicians have an incentive to focus on narrower groups tend to have higher levels of agricultural support (see also Olper and Raimondi, 2004). Moreover, Klomp and De Haan (2013) find that in industrial countries the election effect on agricultural support is stronger under majoritarian than under proportional electoral systems. In contrast, in developing countries the election effect is stronger under proportional electoral systems. None of the electoral systems present in the different CEE countries can be regarded as a pure majoritarian system. Therefore, it is not directly clear which of the two situations is the most appropriate for the CEE countries (Birch, 2001). To capture the different dimensions of the political system, we create dummies for mixed vs. proportional and for parliamentary vs. presidential political systems. We base our political system variables on information reported in the World Bank’s Database of Political Institutions and Birch (2001). Countries are classified as follows: If the president has no legislative powers in the realm of agricultural policy and the government is accountable to parliament through a confidence requirement, the country is classified as a parliamentary regime; otherwise it is classified as a presidential system.14 Likewise, there 11 Agricultural physical capital is defined as the sum of machinery, equipment and fixed livestock. 12 We also used the share of the work force active in the agricultural sector. However, this gave similar results. 13 See also Olper et al. (2014) which report a significant positive effect of a democratic transition that ranges between 5% and 10% using the rate of assistance as their dependent variable. 14 About 60% of the elections included in our sample are in a strict proportional system, while 40% use a mixed system. Besides, more than 55% of the elections are in a presidential system.

may be differences between parliamentary vs. presidential systems. In contrast to a parliamentary system, in a presidential system the executive cannot be brought down by the legislature, but it is directly accountable to the voters. Thus, legislators have weaker incentives to vote according to party or coalition lines.15

Main results This section reports the results for political cycles in agricultural liberalization in Central and Eastern Europe. In view of the unequal distribution of the availability of the data across the countries, we clustered the Huber–White standard errors obtained by using the jackknife approach with 1000 replications. As already mentioned above, the consistency of the GMM estimator depends on the validity of the instruments. To address this issue we consider two specification tests. The first is Sargan’s test of over-identifying restrictions, which tests the overall validity of the instruments by analyzing the sample analog of the moment conditions used in the estimation process. The second test examines the hypothesis that the error term is not serially correlated. The Sargan test provides no evidence of misspecification, while the serial correlation tests point to first- but no second-order autocorrelation of the residuals, which is in accordance with the assumptions underlying the selection of instruments. Alternatively, we used the Hansen test to explorer the validity of our instruments used. The p-values of the Hansen over-identifying restriction test indicate also that the GMM instruments are valid. Table 1 shows our first results. In the first part we add our election cycle measure into the baseline specification, including the control variables outlined in the previous section. The results indicate that in an election year the change in prices and markets liberalization is about 2.4 percentage-points lower than in nonelection years.16 The economic significance is rather moderate since liberalization increases by more than 12% during an average cabinet term. That is, reforms are only slowed down during an election year. This result provide some evidence that incumbent governments expect that voters are short-sighted. However, as already outlined above, the preferred policies by farmers may differ depending upon whether they are being taxed or if they receive subsidies in the pre-reform situation. In both cases, the degree of price and market liberalization is low, but having different policy implications. So to have a closer look on this latter issue, we include an interaction term between our election indicator and a Commonwealth of Independent States dummy. The agricultural sector in these countries is heavily taxed and dominated by state ownership. The results in column (2) confirm our view that the impact of elections is much more complex. In non-CIS countries liberalization in prices and markets is slowed down or reversed in an election year, while in CIS countries a more liberalized market is promoted in an election year. One explanation is that farmers in non-CIS countries are an important constituency which benefit from such reversal, while in CIS countries farmers benefit from more liberalized markets. These results suggest that the incumbent government uses agricultural reforms in prices and markets as an instrument to win votes for its re-election. Since this dimension is the most comparable with the in the literature commonly used Nominal Rate of Assistance (NRA), our results are in line with Klomp and De Haan (2013), Thies and Porche (2007) and Park and Jensen (2007). These studies find all a significant

15 We have included the real GDP per capita, trade openness and land and capital endowment using the natural logarithm to make them more normal distributed. 16 The complete election effect is the sum of the pre- and election year impact. For instance when the elections are held in March, than the election effect on the prices 3  9 and markets dimension would be 12  0:024 þ 12  0:024  100 ¼ 2:4%.

J. Klomp / Food Policy 49 (2014) 332–346

positive election effect on agricultural income support or trade protection.17 Furthermore, the results in columns (3) and (4) demonstrate that land market privatization is spurred during an election year. However, the impact of these reforms does not differ between CIS and non-CIS countries. In both country samples, farmers may benefit from more private ownership as already mentioned above. During an election year the index of land market privatization is about 1% higher than in a non-election year. Although statistical significant, the economic significance seems again rather small. Moreover, the results on privatization of the agro-process industry indicate that they are not affected by upcoming elections. One explanation is that farmers may lobby for a more open agroprocess industry, while the industry itself may lobby for less rigorous reforms or even for more entry barriers. These two opposite effects appear to cancel out each other. Besides, before the gains are realized for farmers of these kind of reforms, for instance, by attracting more foreign competitors, is a long-lasting process which makes it a less attractive instrument to use with upcoming elections. A similar story holds for the reforms in rural finance. We find no significant impact of elections on these kind of reforms since some farmers may benefit from lower interest rate due to more competition, whereas others may be hurt as they benefit from interest rate cuts subsidized by the government or from their ties with the financial sector. Alternatively, the gains for the average farmer may be only realized in the long run when bank competition is increased. Therefore are these reforms in the short run ineffective and unsuitable for the incumbent government to increase the likelihood of re-election. Finally, not so surprising, we find no statistical significant election effect on reforms in the institutional framework since these reforms move gradually due to the political process. These kind of reforms are therefore less appropriate to use as a reelection strategy. According to Rozelle and Swinnen (2004) political lobbying and bureaucratic connections even made that institutional market reforms were discouraged. The results presented so far give some empirical support for our view that governments may target only specific dimensions of agricultural liberalization in an election year to benefit farmers. In detail, although the economic significance is rather small, we find some empirical evidence for an election effect in reforms on prices and markets and on land markets. One explanation is that these policies may affect the economic wellbeing of farmers the most directly and the gains of these policies are realized the quickest. In the bottom part of Table 2 we add our partisan cycle and nationalistic government indicator. Based on these regression results, we can draw a number of conclusions. First, the coefficient on the partisan cycle on prices and markets liberalization has a positive sign and turns out to be statistically significant at common confidence levels. That is, left-wing governments are less protective to the agricultural sector than right-wing governments. A full right-wing government reduces prices and markets liberalization by about 0.8 percentage-points annually compared to a cabinet with only left-wing parties.18 This result is in line with the view that right-wing governments take the interests of their agricultural constituency into account. However, our results again depend on the institutional origin of a country. In CIS countries right-wing governments increase price and market liberalization as in most of these 17 As our data stops at the moment a country enters the EU, we are not able to capture the impact of election cycles on tariff protection and CAP subsidies. To complete the picture and assess the robustness of the ‘prices and markets liberalization’ index, we use the Nominal Rate of Assistance reported by Anderson and Valenzuela (2008), which can be considered to be the inverse of the price and market liberalization index. The results using the NRA between 1995 and 2009 point in the same direction as the results by using the price and market liberalization index (results are available upon request). 18 Calculated as ð5  1Þ  0:002  100 ¼ 0:8%.

339

countries agriculture is initially taxed, which is then reduced by right-wing parties. In non-CIS countries, right-wing countries support the agricultural sector by supplying subsidies. Both these results indicate that right-wing governments use prices and markets reforms to increase the benefits for farmer. Second, right-wing governments implement more reforms to privatize the land market and the agro-processing industry compared to left-wing governments. This in line with the standard partisan theory arguing that right-wing governments favor less state control and more private ownership. According to Bjørnskov and Potrafke (2011) right-wing political parties are more in favor of the so-called Washington Consensus approach of a rapid transition. Meaning, excessively slow reforms might give special interests time to regroup, while a fast transition would lead more quicker to a free market system. Conversely, the left-wing political parties are more associated with the gradualist approach favouring a relatively slow reform pace and only gradual liberalization that maintains some state enterprises and government regulation to minimize the social costs of transition or distributing the social costs across a longer period of time. This latter implies a higher degree of respect for existing power structures and a continuation of political control over the economy. In contrast, we find any evidence of a significant partisan cycle on the rural finance system or institutional framework. This result may be explained by the fact that political preferences might be less pronounced in reforming the rural financial system, while reforming the institution environment is rather hard due to the political process and the number of veto players. Finally, we find that nationalistic governments lower the speed of more market-orientated reforms significantly as they are in favor of the old system of institutions and collective way of farming. As a consequence these type of regimes often still use state control and ownership to reduce competition. These results are in line with the empirical conclusions from Bjørnskov and Potrafke (2011) on privatization in the business sector in CEE countries which is also retarded by nationalistic cabinets. Robustness analysis To sum up the results presented so far, we provide some support that political cycles influence certain reforms in the agricultural sector in CEE countries. However, the impact and significance may differ across some elements of the political system in place. For instance, Olper et al. (2010) explore how democratic transitions into different constitutional systems affect agricultural protection. Their results strongly support the notion that a shift from autocracy or a majoritarian democracy to a proportional democracy induces a strong increase in agricultural protection. A similar but weaker effect was detected for transition to a presidential system.19 However, as already pointed out by Olper and Raimondi (2013) some elements of the political system may be endogenous (see also Aghion et al., 2004). That is, different public policies may generate different benefits for the various voter groups present within a country. Meaning, voters differ in their preferences over electoral rules and politicians take this into account in times of constitutional changes. For instance, Ticchi and Vindigni (2010) demonstrate that a democracy is more difficult to sustain politically in a more unequal society since greater inequality tends to undermine the stability of the coalition supporting it. As argued by Persson and Tabellini (2003) endogeneity issues of electoral rules and forms of government on policy outcomes are mainly due to problems of selection bias. In our sample with CEE countries with 19 Other studies examining the impact of political institutions on agricultural protection include Olper and Raimondi (2004, 2010, 2013) and Henning (2008), Henning and Struve (2007) and Henning et al. (2011).

340

Table 2 Political cycles and agricultural liberalization. Dependent variable: D ln agrilib Land market reform

Privatization of agro-processing

Rural finance systems

Institutional framework

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

0.024 [1.93]*

0.031 [2.03]** 0.046 [1.71]*

0.010 [1.87]*

0.010 [1.79]* 0.002 [1.01]

0.006 [1.33]

0.005 [1.29] 0.002 [1.44]

0.010 [1.32]

0.009 [1.08] 0.001 [1.32]

0.006 [1.06]

0.005 [1.08] 0.002 [1.08]

Controls Country-specific time trend Observations Number of countries Arellano–Bond test AR(1) p-value) Arellano–Bond test AR(2) Sargan test (p-value) Hansen difference test (p-value)

YES YES 121 19 0.000 0.696 0.742 0.421

YES YES 121 19 0.000 0.669 0.735 0.388

YES YES 121 19 0.000 0.694 0.701 0.357

YES YES 121 19 0.000 0.723 0.745 0.403

YES YES 121 19 0.000 0.651 0.733 0.375

YES YES 121 19 0.000 0.726 0.698 0.378

YES YES 121 19 0.000 0.683 0.703 0.382

YES YES 121 19 0.000 0.655 0.676 0.359

YES YES 121 19 0.000 0.664 0.715 0.356

YES YES 121 19 0.000 0.726 0.678 0.411

Partisan cycle

0.003 [2.01]**

0.002 [2.02]** 0.003 [1.94]*

0.003 [1.71]*

0.003 [1.76]* 0.004 [1.01]

0.002 [1.99]**

0.002 [1.83]* 0.001 [1.24]

0.003 [1.30]

0.003 [1.59] 0.001 [1.11]

0.003 [1.05]

0.002 [1.60] 0.002 [1.08]

0.003 [2.08]**

0.004 [1.97]** 0.002 [1.23]

0.002 [2.02]**

0.002 [2.01]** 0.001 [1.45]

0.003 [2.17]**

0.004 [2.30]** 0.002 [1.34]

0.003 [1.08]

0.003 [1.13] 0.001 [1.02]

0.003 [1.57]

0.003 [1.42] 0.001 [1.11]

YES YES 121 19 0.000 0.651 0.743 0.494

YES YES 121 19 0.000 0.683 0.735 0.524

YES YES 121 19 0.000 0.669 0.730 0.483

YES YES 121 19 0.000 0.667 0.662 0.491

YES YES 121 19 0.000 0.668 0.659 0.480

YES YES 121 19 0.000 0.704 0.686 0.504

YES YES 121 19 0.000 0.657 0.717 0.499

YES YES 121 19 0.000 0.655 0.720 0.491

YES YES 121 19 0.000 0.728 0.736 0.453

YES YES 121 19 0.000 0.699 0.678 0.487

Election cycle Election cycle  CIS countries

Partisan cycle  post-communistic Nationalistic governments Nationalistic gov.  CIS countries Controls Country-specific time trend Observations Number of countries Arellano–Bond test AR(1) p-value) Arellano–Bond test AR(2) p-value) Sargan test (p-value) Hansen difference test (p-value)

Note: This table shows the empirical results of estimating Eq. (2) using system-GMM technique. The top part of the table reports the results on the election cycle, while the bottom part gives the main results on the partisan cycle. We estimate the models including GDP per capita, growth rate of GDP, inflation, trade openness, size agricultural sector, rural population share, land endowment, agricultural export, capital endowment, level of democracy, mixed electoral system, parliamentary system, IMF program dummy, years to EU membership as control variables. Robust t-values are shown in parentheses. * Indicate significance at 10%. ** Indicate significance at 5%.

J. Klomp / Food Policy 49 (2014) 332–346

Prices and markets

Table 3 Election cycles, political system and agricultural liberalization. Dependent variable: D ln agrilib Land market reform

Privatization of agro-processing

Rural finance systems

Institutional framework

Non-CIS (1)

CIS (2)

Non-CIS (3)

CIS (4)

Non-CIS (5)

CIS (6)

Non-CIS (7)

CIS (8)

Non-CIS (9)

CIS (10)

0.030 [1.95]* 0.027 [1.80]*

0.013 [1.80]* 0.015 [1.67]*

0.010 [1.85]* 0.009 [1.80]*

0.008 [1.73]* 0.007 [1.70]*

0.009 [1.44] 0.009 [1.30]

0.011 [1.49] 0.011 [1.29]

0.006 [0.91] 0.005 [1.08]

0.006 [1.12] 0.006 [1.14]

0.009 [1.60] 0.011 [1.54]

0.008 [1.45] 0.009 [1.32]

Difference

ns

ns

ns

ns

ns

ns

ns

ns

ns

ns

Controls Country-specific time trend Observations Number of countries Arellano–Bond test AR(1) p-value) Arellano–Bond test AR(2) p-value) Sargan test (p-value) Hansen difference test (p-value)

YES YES 121 19 0.000 0.721 0.731 0.681

YES YES 121 19 0.000 0.727 0.722 0.658

YES YES 121 19 0.000 0.719 0.733 0.749

YES YES 121 19 0.000 0.669 0.743 0.651

YES YES 121 19 0.000 0.713 0.652 0.686

YES YES 121 19 0.000 0.725 0.699 0.706

YES YES 121 19 0.000 0.668 0.656 0.699

YES YES 121 19 0.000 0.720 0.707 0.745

YES YES 121 19 0.000 0.703 0.698 0.673

YES YES 121 19 0.000 0.720 0.689 0.713

(3)

Mixed election

(4)

Proportional elections

0.045 [1.97]** 0.020 [1.77]*

0.021 [1.96]* 0.012 [1.76]*

0.011 [1.84]* 0.010 [1.78]*

0.010 [1.79]* 0.009 [1.70]*

0.010 [1.43] 0.009 [1.24]

0.006 [1.37] 0.007 [1.04]

0.009 [1.45] 0.008 [0.85]

0.008 [1.24] 0.007 [0.86]

0.010 [1.23] 0.009 [1.13]

0.011 [1.60] 0.009 [1.42]

Election cycle (1) Established democracy (2)

Transition democracy

Difference

sig

sig

ns

ns

ns

ns

ns

ns

ns

ns

(5)

Presidential

(6)

Parliamentary

0.033* [1.94] 0.036 [1.98]*

0.012 [1.79]* 0.015 [1.76]*

0.007 [1.81]* 0.009 [1.76]*

0.009 [1.74]* 0.011 [1.69]*

0.009 [1.40] 0.009 [1.49]

0.012 [1.55] 0.011 [1.47]

0.005 [0.89] 0.006 [1.06]

0.007 [1.12] 0.007 [1.25]

0.009 [1.54] 0.009 [1.53]

0.010 [1.45] 0.011 [1.46]

Difference

ns

ns

ns

ns

ns

ns

ns

ns

ns

ns

Controls Country-specific time trend Observations Number of countries Arellano–Bond test AR(1) p-value) Arellano–Bond test AR(2) p-value) Sargan test (p-value) Hansen difference test (p-value)

YES YES 127 20 0.000 0.670 0.739 0.655

YES YES 127 20 0.000 0.693 0.670 0.725

YES YES 127 20 0.000 0.727 0.706 0.740

YES YES 127 20 0.000 0.657 0.698 0.655

YES YES 127 20 0.000 0.694 0.698 0.675

YES YES 127 20 0.000 0.694 0.747 0.726

YES YES 127 20 0.000 0.705 0.697 0.672

YES YES 127 20 0.000 0.742 0.660 0.723

YES YES 127 20 0.000 0.659 0.716 0.700

YES YES 127 20 0.000 0.665 0.743 0.695

All elections

0.019 [1.82]*

0.017 [1.77]*

0.008 [1.74]*

0.006 [1.64]

0.007 [1.09]

0.008 [1.26]

0.005 [0.88]

0.005 [1.0]

0.006 [1.34]

0.008 [1.09]

Controls Country-specific time trend Observations Number of countries Arellano–Bond test AR(1) p-value) Arellano–Bond test AR(2) p-value) Sargan test (p-value) Hansen difference test (p-value)

YES YES 127 20 0.000 0.689 0.665 0.664

YES YES 127 20 0.000 0.668 0.651 0.747

YES YES 127 20 0.000 0.705 0.717 0.689

YES YES 127 20 0.000 0.658 0.683 0.651

YES YES 127 20 0.000 0.704 0.697 0.732

YES YES 127 20 0.000 0.718 0.685 0.697

YES YES 127 20 0.000 0.700 0.721 0.651

YES YES 127 20 0.000 0.653 0.726 0.685

YES YES 127 20 0.000 0.685 0.689 0.654

YES YES 127 20 0.000 0.719 0.665 0.659

(8)

Pre-election dummy

(9)

Post-election dummy

0.028 [1.91]* 0.004 [1.67]*

0.019 [1.81]* 0.002 [1.44]

0.013 [1.83]* 0.006 [1.30]

0.010 [1.74]* 0.008 [1.11]

0.016 [1.38] 0.008 [1.14]

0.013 [1.52] 0.007 [1.19]

0.008 [1.02] 0.004 [0.98]

0.008 [1.28] 0.006 [0.97]

0.015 [1.25] 0.009 [1.36]

0.012 [1.35] 0.006 [1.46]

(7)

341

(continued on next page)

J. Klomp / Food Policy 49 (2014) 332–346

Prices and markets

Note: This table shows the empirical results of estimating Eq. (6) including the election cycle and using the system-GMM technique. We estimate the models including GDP per capita, growth rate of GDP, inflation, trade openness, size agricultural sector, rural population share, land endowment, agricultural export, capital endowment, level of democracy, mixed electoral system, parliamentary system, IMF program dummy, years to EU membership as control variables. Robust t-values are shown in parentheses. ‘‘sig’’ refers to a significant difference between the two sample, while ‘‘ns’’ indicates an insignificant difference. * Indicate significance at 10%. ** Indicate significance at 5%.

YES YES 121 19 0.000 0.715 0.696 0.656 YES YES 121 19 0.000 0.706 0.699 0.669 YES YES 121 19 0.000 0.656 0.740 0.692 YES YES 121 19 0.000 0.663 0.746 0.721 YES YES 121 19 0.000 0.726 0.693 0.709 YES YES 121 19 0.000 0.699 0.702 0.671 YES YES 121 19 0.000 0.732 0.713 0.686 YES YES 121 19 0.000 0.656 0.716 0.687 YES YES 121 19 0.000 0.722 0.735 0.727 Controls Country-specific time trend Observations Number of countries Arellano–Bond test AR(1) p-value) Arellano–Bond test AR(2) p-value) Sargan test (p-value) Hansen difference test (p-value)

CIS (2)

YES YES 121 19 0.000 0.700 0.714 0.706

CIS (10) CIS (8)

Rural finance systems

Non-CIS (7) CIS (6)

Privatization of agro-processing

Non-CIS (3) Non-CIS (1)

CIS (4)

Land market reform Prices and markets

Dependent variable: D ln agrilib Table 3 (continued)

Non-CIS (5)

Institutional framework

J. Klomp / Food Policy 49 (2014) 332–346

Non-CIS (9)

342

only a short democratic history, many political institutions are still shaped by their communistic origin. The last twenty years a number democratic transitions took place in these countries, making them to re-organize their political system. Therefore, we cannot complete rule out that there is an overlap between this transition process and the impact of political driven cycles on economic outcomes. The next step in our analysis is to examine whether there is a conditional effect of the different elements of the political system and the political cycles using interaction terms. We estimate the following model:

ln agrilibit  ln agrilibit1 ¼ ait þ ðc  1Þ ln agrilibit1 þ

J X bk xkit1 j¼1

þ l cycleit þ u systemit þ g ðcycleit  systemit Þ þ eit

ð6Þ

where cycleit refers to the election or ideology indicator, while systemit represents different dimensions of the political regime in place. The conditional effect of government ideology on the effect of elections can be calculated by the derivation of Eq. (6) with respect to cycle.

@ D ln agrilib ¼ l þ g system @cycle The statistical significance of the interaction effects cannot be tested with a simple t-test on the coefficient of the interaction terms but must be based on the estimated cross-partial derivative (Green, 2002). The standard error of interest is

r^ @sD ln agrilib ¼

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi varðlÞ þ system2  varðgÞ þ 2  system  covðl; gÞ

@cycle

We consider various elements of the political system on which our political cycle variable may be condition. As already shown in the previous section, it is important to distinguish between CIS countries and non-CIS countries as they differ on the type of government intervention, such as taxation versus subsidizing of the agricultural production or the political institutions present. First, according to Brender and Drazen (2005) the magnitude of the election effect declines with the level of democracy and experience with the electoral system. To capture this notion we first create a dummy variable taking the value one for countries with a continuous democratic history of more than 10 years since 1989, which is about equal to 3 presidential or parliamentary elections. As before, we count each country-year with a Polity score larger or equal to 5 as a democratic year. The hypothesis tested is that the longer the democratic history, elections will have less impact on public policy as voters have become aware about public policy manipulation before an election.20 The results in row (1) and (2) of Tables 3 and 4 indicate that the impact of elections or government ideology does not systematically differ between the two samples. Thus, political cycles do not depend on the democratic history. This results contradicts the conclusions by Klomp and De Haan (2013) who find that agricultural support in an election year is larger in countries with a short democratic history. One explanation for this different outcome is that the democratic history of most CEEC’s is still relative short. Next, as already mentioned above, there are major differences in the provision of public goods between majoritarian vs. proportional systems and between parliamentary vs. presidential systems (see Persson and Tabellini, 2000, 2002, 2003; Grossman and Helpman, 2005; Olper and Raimondi, 2010, 2011, 2013). In a majoritarian system an electoral district is generally small and the politician who wins the majority of the votes represents this district in parliament. Such a system gives politicians a strong incentive to target policies towards a particular constituency. In 20 We also used the Polity IV score of 4, although the dataset is then drastically reduced, the results remain in line.

Table 4 Partisan cycles, political system and agricultural liberalization. Dependent variable: D ln agrilib

Partisan cycle (1) Established democracy (2)

Transition democracy

Prices and markets

Land market reform

Privatization of agro-processing

Rural finance systems

Institutional framework

Non-CIS (1)

CIS (2)

Non-CIS (3)

CIS (4)

Non-CIS (5)

CIS (6)

Non-CIS (7)

CIS (8)

Non-CIS (9)

CIS (10)

0.002 [1.97]** 0.002 [1.93]*

0.002 [1.94]* 0.002 [1.87]*

0.002 [1.89]* 0.002 [1.76]*

0.002 [1.77]* 0.002 [1.67]*

0.003 [1.93]* 0.003 [1.89]*

0.003 [1.73]* 0.003 [1.86]*

0.003 [1.15] 0.003 [1.29]

1.225 [1.16] 1.144 [0.99]

0.003 [1.76] 0.004 [1.71]

0.001 [1.52] 0.001 [1.48]

0.43

0.33

0.34

0.60

0.39

0.45

0.64

0.49

0.63

0.48

Controls Country-specific time trend Observations Number of countries Arellano–Bond test AR(1) p-value) Arellano–Bond test AR(2) p-value) Sargan test (p-value) Hansen difference test (p-value)

YES YES 201 19 0.000 0.740 0.720 0.714

YES YES 201 19 0.000 0.747 0.656 0.676

YES YES 201 19 0.000 0.745 0.651 0.659

YES YES 201 19 0.000 0.677 0.735 0.706

YES YES 201 19 0.000 0.721 0.747 0.750

YES YES 201 19 0.000 0.654 0.689 0.738

YES YES 201 19 0.000 0.670 0.720 0.677

YES YES 201 19 0.000 0.673 0.680 0.707

YES YES 201 19 0.000 0.736 0.710 0.688

YES YES 201 19 0.000 0.747 0.748 0.664

(3)

Proportional elections

0.003 [1.93]*

0.003 [1.91]*

0.003 [1.93]*

0.002 [1.82]*

0.004 [1.91]*

0.005 [1.75]*

0.004 [1.33]

1.672 [1.25]

0.005 [1.63]

0.002 [1.53]

(4)

Mixed elections

0.002 [1.86]*

0.001 [1.80]*

0.002 [1.83]*

0.001 [1.89]*

0.002 [1.77]*

0.003 [1.67]*

0.002 [1.02]

0.942 [0.93]

0.003 [1.33]

0.001 [1.19]

P-value difference

0.47

0.32

0.38

0.34

0.56

0.58

0.54

0.50

0.34

0.38

(5)

Presidential

(6)

Parliamentary

0.002 [1.97]** 0.002 [1.92]*

0.002 [1.86]* 0.002 [1.81]*

0.002 [1.89]* 0.002 [1.72]*

0.002 [1.80]* 0.002 [1.73]*

0.003 [1.85]* 0.003 [1.87]*

0.004 [1.84]* 0.004 [1.93]*

0.003 [1.14] 0.003 [1.09]

1.152 [1.02] 1.255 [1.18]

0.004 [1.43] 0.003 [1.38]

0.001 [1.33] 0.002 [1.43]

P-value difference

0.52

0.56

0.32

0.45

0.56

0.64

0.34

0.52

0.30

0.37

Controls Country-specific time trend Observations Number of countries Arellano–Bond test AR(1) p-value) Arellano–Bond test AR(2) p-value) Sargan test (p-value) Hansen difference test (p-value)

YES YES 201 19 0.000 0.743 0.744 0.381

YES YES 201 19 0.000 0.655 0.737 0.437

YES YES 201 19 0.000 0.686 0.718 0.436

YES YES 201 19 0.000 0.659 0.660 0.400

YES YES 201 19 0.000 0.733 0.698 0.417

YES YES 201 19 0.000 0.748 0.731 0.447

YES YES 201 19 0.000 0.715 0.652 0.373

YES YES 201 19 0.000 0.673 0.721 0.387

YES YES 201 19 0.000 0.725 0.651 0.450

YES YES 201 19 0.000 0.714 0.674 0.428

J. Klomp / Food Policy 49 (2014) 332–346

P-value difference

Note: This table shows the empirical results of estimating Eq. (6) including the government ideology variables and using the system-GMM technique. We estimate the models including GDP per capita, growth rate of GDP, inflation, trade openness, size agricultural sector, rural population share, land endowment, agricultural export, capital endowment, level of democracy, mixed electoral system, parliamentary system, IMF program dummy, years to EU membership as control variables. Robust t-values are shown in parentheses. * Indicates significance at 10%. ** Indicates significance at 5%.

343

344

J. Klomp / Food Policy 49 (2014) 332–346

proportional systems, public policies are arguably more directed towards programs benefiting large groups in the population. As to differences between presidential vs. parliamentary systems, the former systems are characterized by separate and direct elections for both the executive and the legislature. In parliamentary systems, the executive is only indirectly formed though the legislature. That is, in a presidential regime, the president is better able to target particular constituencies, while in a parliamentary system general interests are better represented. Sharper conflict of interests in presidential systems should induce greater preference for targeted instruments, such as agricultural support, to powerful minorities. Most CEE countries are in an economic transition, while the share of the agricultural sector to the economy is still large, it has declined rapidly the last decade. For instance, the share of GDP contributed by the agricultural sector dropped from more than 20% in 1996 to about 11% in 2007. To examine if political cycles are driven by the electoral rules in place, we estimate the baseline economic specification including interaction effects between the political cycle and the indicators of the political system (mixed vs. proportional systems, and parliamentary vs. presidential systems). The results as shown in the middle part of Tables 3 and 4 suggest that mixed elections are weakly associated with higher pre-election manipulation in prices and markets liberalization. On the other dimensions we find no conditional election impact. Moreover, we also do not find any evidence that right-wing parties in mixed systems are less inclined to liberalize the agricultural sector than right-wing parties in proportional systems. In addition, the influence of election and partisan cycles on liberalization in the agricultural sector does not differ systematically across presidential and parliamentary regimes. In our empirical strategy so far, we only include elections which are held on a fixed date or are announced at least one year in advance. We restrict the analysis to these elections because in endogenous elections within one year the possibilities to use public policy for election purposes is limited. However, we have also estimated the models including all elections. The results in the bottom part of Table 3 are fairly similar to those reported in Table 2 suggesting that endogenous elections are only a minor issue. Finally, we have used an alternative measures for election cycles. Several studies use dummy variables for recognizing a pre-election and post-election year in their econometric specification (Shi and Svensson, 2006; Brender and Drazen, 2005). As a robustness check we re-estimate the main model including these two dummies. The results (shown in the bottom part of Table 3) on the pre-election year dummy are similar to the results obtained above. However, in non-CIS countries newly elected governments are more inclined to implement price and market reforms in a post-election year. One rational explanation is that in order to obtain all the gains for the society at large during the legislature, politician have to implement the liberalization reform immediately after elections to allow as much time as possible for the gains from the reform and to produce clear winners before voters cast their ballots in the next election (Haggard and Webb, 1993; Williamson, 1994). Conclusions Since the beginning of the 90s, many Central and Eastern European countries have provided liberalization reforms to their agricultural sector, such as reducing tariffs and cutting subsidies. Politicians may use agricultural policies for economic motives, but also for political purposes. In this study, we examine whether election cycles and the government ideology affect the multiple dimensions of agricultural liberalization in CEE countries. To explore this relationship, we use a dynamic panel model including about 20 Central and Eastern European countries over the period 1996–2005. We use different dimensions of liberalization reported

by the World Bank which refer to: (1) prices and markets liberalization; (2) land market privatization; (3) agro-processing and input supply privatization; (4) rural finance reforms and (5) market institutions liberalization. After testing for the sensitivity of the results, we can draw a number of conclusions. First, liberalization in prices and markets is slowed down or partly reversed under the influence of upcoming elections in countries which subsidize the agricultural sector or have open markets. However, in countries where the agricultural sector previously was taxed or under state control, there we find a positive election impact on prices and markets reforms. This latter result can be explained that in these countries, farmers benefit from a more open trade system with less distortions. Second, governments stimulate land market reforms during an election year by allowing for more private ownership. Third, we do not find any evidence of electoral cycles in reforms in the agro-processing industry, rural finance and institutions as reforming these dimensions at a short notice is politically difficult or the individual benefits for farmers may be too small to signal competence of the cabinet. Consequently, reforms affecting these dimensions are less appropriate as a re-election instrument. Fourth, right-wing governments support or protect the interests of the agricultural sector more than left-wing governments. To be more precise, Former Soviet Union member states ruled by a right-wing cabinet open up their markets more often and decline the agricultural taxation, while in non-former Soviet Union member states right-wing governments start subsidizing farmers more. Fifth, nationalist-led governments are associated with slower transition speeds to re-enforce state control. Finally, it turns out that the effect of election cycles in price and market liberalization is partly conditional on the political system in place. In particular, the election effect is stronger under a mixed electoral system compared to a proportional system. Our interpretation of this finding is that a mixed system gives the incumbent a stronger incentive to target transfers to particular interest groups, like agricultural producers than under a strict proportional system. One limitation in our study is that since many CEEC’s have gone through some major economic and political transition during our period of analysis, we cannot completely rule out that our results are affected by this process. Besides, as we focus on competitive elections, we only consider democratic country-years, which reduces our dataset as many CEEC’s have still a short democratic history. Appendix A See Tables A1 and A2. Table A1 Countries included in the empirical analysis. Albania Armenia Bulgaria Croatia Czech Republic Estonia Georgia Hungary Kyrgyz Republic Latvia Lithuania Macedonia, FYR Moldova Poland Romania Russian Federation Serbia Slovak Republic Slovenia Ukraine

345

J. Klomp / Food Policy 49 (2014) 332–346 Table A2 Data sources used. Variable

Description

Source

Real GDP per capita Economic growth Inflation Trade openness Size agricultural sector Rural population share Land endowment

Real GDP per capita in constant US dollars of 2005 taken in logarithms

Heston et al. (2011)

Growth rate of real GDP per capita Calculated by p/(1 + p) where p is the change in the GDP deflator Sum of import and export as a share of total GDP Share of GDP contributed by the agricultural sector.

Agricultural export Capital endowment Level of democracy Mixed electoral system Parliamentary system IMF program

Export of agricultural products and commodities as a share of total exports

Heston et al. (2011) Heston et al. (2011) Heston et al. (2011) World Bank (2011) and FAO (2013) World Bank (2011) and FAO (2013) World Bank (2011) and FAO (2013) FAO (2013)

Total of agricultural capital (machinery, equipment and fixed livestock) divided by the number of workers in the agricultural sector taken in logarithms Polity IV score

World Bank (2011) and FAO (2013) Jaggers et al. (2011)

Dummy variable that is one if the election is in a mixed electoral system, zero otherwise

Years to EU membership

The number of years before a country becomes an official member of the EU

Update of Beck et al. (2001), Birch (2001) Update of Beck et al. (2001), Birch (2001) Dreher (2006) and www.imf.org www.eu.com

Share of total population living in the rural areas Total hectare of arable land divided by the labor force working in the agricultural sector taken in logarithms

Dummy variable that is one if the election is in a parliamentary system, zero otherwise A dummy variable taking the value one in a country year when a country is subject to an IMF loan, zero otherwise

References Aghion, P., Alesina, A., Trebbi, F., 2004. Endogenous political institutions. Q. J. Econ. 119, 565–611. Anderson, K., 1995. Lobbying incentives and the pattern of protection in rich and poor countries. Econ. Dev. Cult. Change 43, 401–423. Anderson, K., Hayami, Y., 1986. The Political Economy of Agricultural Protection. Allen and Un win, Sydney. Anderson, K., Valenzuela, E., 2008. Estimates of Global Distortions to Agricultural Incentives, 1955 to 2007. World Bank, Washington, DC. Anderson, K., Rausser, G., Swinnen, J.F.M., 2013. Political economy of public policies: insights from distortions to agricultural and food markets. J. Econ. Lit. 51, 423– 477. 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. Arellano, M., Bover, O., 1995. Another look at the instrumental variable estimation of error-components models. J. Econom. 68, 29–51. Bates, R., 1989. Beyond the Miracle of the Market, The Political Economy of Agrarian Development in Kenya. Cambridge University Press, Cambridge. Bates, R., Block, S., 2011a. Political institutions and agricultural trade interventions in Africa. Am. J. Agric. Econ. 93, 317–323. Bates, R., Block, S., 2011b. Agricultural trade interventions in Africa. pp. 304–331. In: Aderson, K. (Ed.), The Political Economy of Agricultural Price Distortions. Cambridge University Press. Beck, T., Clarke, G., Groff, A., Keefer, P., Walsh, P., 2001. New tools in comparative political. Economy, The database of political institutions. World Bank Econ. Rev. 15, 165–176. Birch, S., 2001. Electoral systems and party systems in Europe East and West. Perspect. Eur. Polit. Soc. 2, 355–377. Bjørnskov, C., Potrafke, N., 2011. Politics and privatization in Central and Eastern Europe. Econ. Transit. 19, 201–230. Blundell, R., Bond, S., 1998. Initial conditions and moment restrictions in dynamic panel models. J. Econ. 87, 115–143. Bortolotti, B., Pinotti, P., 2008. Delayed privatization. Public Choice 136, 331–351. Bortolotti, B., Fantini, M., Siniscalco, D., 2001. Privatisation, politics, institutions, and financial markets. Emergy Market Rev. 2, 109–137. Bortolotti, B., Fantini, M., Siniscalco, D., 2003. Privatisation around the world, evidence from panel data. J. Public Econ. 88, 305–332. Brender, A., Drazen, A., 2005. Political budget cycles in new versus established democracies. J. Monetary Econ. 52, 1271–1295. Buchenrieder, G., Hanf, J., Pieniadz, A., 2009. 20 years of transition in the agri-food sector. Ger. J. Agric. Econ. 58, 285–293. Ciaian, P., Swinnen, J., 2007. Distortions to agricultural incentives in Central and Eastern Europe. Agricultural Distortions Working Paper 07. Washington. Csaki, G., 2000. Agricultural reforms in Central and Eastern Europe and the former Soviet Union. Agric. Econ. 22, 37–54. Cungu, A., Swinnen, J., 1999. Albania’s radical agrarian reform. Econ. Dev. Cult. Change 47, 605–619.

Deininger, K., 2003. Land markets in developing and transition economies: impact of liberalization and implications for future reform. Am. J. Agric. Econ. 85, 1217– 1222. Dreher, A., 2006. IMF and economic growth, the effects of programs, loans, and compliance with conditionality. World Dev. 34, 769–788. Falkowski, J., Olper, A., 2012. Political Competition and Support for Agriculture. University of Warsaw, Faculty of Economic Sciences, Working Papers 18. FAO, 2013. FAOSTAT. faostat.fao.org. Fidrmuc, J., 2000. Economics of voting in post-communist countries. Elect. Stud. 19, 199–217. Franzese, R., 2000. Electoral and partisan manipulation of public debt in developed democracies, 1956–1990. In: Strauch, R., von Hagen, J. (Eds.), Institutions, Politics and Fiscal Policy. Kluwer Academic Press, Dordrecht, pp. 61–83. Frye, T., Mansfield, E., 2004. Timing is everything: election and trade liberalization in the post communist world. Comp. Polit. Stud. 37, 371–398. Giuliano, P., Mishra, P., Spilimberg, A., 2013. Democracy and reforms: evidence from a new dataset. Am. Econ. J.: Macroecon. 5, 179–204. Green, W., 2002. Econometric Analysis. Prentice Hall, New Jersey. Grossman, G., Helpman, E., 2005. A protectionist bias in majoritarian politics. Quart. J. Econ. 120, 1239–1282. Haggard, S., Webb, S., 1993. What do we know about the political economy of economic policy reform? World Bank Policy Res. Obser. 8, 143–167. Henning, C., 2008. Determinants of Agricultural Protection in an International Perspective, The Role of Political Institutions. IFPRI Discussion Paper 00805. Henning, C., Struve, C., 2007. Postelection Bargaining and Special Interest Politics in Parliamentary Systems, The Case of Agricultural Protection. In: Hinich, M., Barnett, W. (Eds.), Topics in Analytical Political Economy. Elsevier, Amsterdam, pp. 45–84. Henning, C., Krampe, E., Aszmann, C., 2011. Electoral Rules and Special Interest Politics. The Case of Agricultural Protection. Paper presented at the 2011 European Public Choice Society Conference. Heston, A., Summers R., Aten, B., 2011. Penn World Table Version 6.3, Center for International Comparisons of Production, Income and Prices at the University of Pennsylvania. Hibbs Jr., D., 1992. Partisan theory after fifteen years. Eur. J. Polit. Econ. 8, 361–373. Jaggers, K., Marshall, M., Gurr, T., 2011. Polity IV: Political regime characteristics and transitions, 1800–2011. Center for Systemic Peace at George Mason University. Johnson, S., 1995.Agriculture and agribusiness reforms in the CEE nations and NIS: Issues and opportunities. Working paper 95-139. Kennedy, L., 1999. Central and Eastern European Agriculture in Transition. In: Columbus, F., (Eds.). Central and Eastern Europe in Transition, Volume 3. Chapter 9. pp. 221–233. Kiewiet, D., 2000. Economic retrospective voting and incentives for policymaking. Elect. Stud. 19, 427–444. Klomp, J., de Haan, J., 2013. Conditional election and partisan cycles in government support to the agricultural sector: an empirical analysis. Am. J. Agric. Econ. 95, 793–818. Knack, S., Keefer, P., 1995. Institutions and economic performance: Cross-country tests using alternative institutional measures. Econ. Polit. 7 (3), 207–227. Lewis-Beck, M., 1977. Explaining Peasant Conservatism, The Western European Case. Brit. J. Polit. Sci. 7, 447–464.

346

J. Klomp / Food Policy 49 (2014) 332–346

Macours, K., Swinnen, J., 2000. Impact of initial conditions and reform policies on agricultural performance in Central and Eastern Europe, the Former Soviet Union, and East Asia. Am. J. Agric. Econ. 82, 1149–1155. Macours, K., Swinnen, J., 2002. Patterns of agrarian transition⁄. Econ. Dev. Cult. Change 50, 365–394. Nordhaus, W., 1975. The political business cycle. Rev. Econ. Stud. 42, 169–190. Olper, A., 2001. Determinants of Agricultural Protection: The Role of Democracy and Institutional Setting Alessandro Olper. J. Agric. Econ. 52, 75–92. Olper, A., 2007. Land Inequality, Government Ideology and Agricultural Protection. Food Policy 32, 67–83. Olper, A., Raimondi, V., 2004. Political Institutions and Milk Policy Outcomes in OECD Countries. In: Van Huylenbroeck, G., Verbeke, W., Lauwers, L. (Eds.), Role of Institutions in Rural Policies and Agricultural Markets. Elsevier, Amsterdam, pp. 153–168. Olper, A., Raimondi, V., 2010. Constitutional Rules and Agricultural Policy Outcomes. In: Anderson, K. (Ed.), The Political Economy of Agricultural Price Distortions. Cambridge University Press, Cambridge, pp. 358–391. Olper, A., Raimondi, V., 2011. Constitutional reforms and food policy. Am. J. Agric. Econ. 93, 324–331. Olper, A., Raimondi, V., 2013. Electoral rules, forms of government and redistributive policy evidence from agriculture and food policy. J. Comp. Econ. 41, 141–158. Olper, A., Raimondi, V., Swinnen, J., 2010. Constitutional Rules and Redistributive Policy, Evidence from Global Agricultural Protection. PUE&PIEC Working Paper No. 10/12. Olper, A., Raimondi, V., Swinnen, J., 2014. Political reforms and public policies, evidence from agricultural protection. World Bank Econ. Rev. 28, 21–47. Olson, M., 1965. The Logic of Collective Action. Harvard University Press, Cambridge. Park, J., Jensen, N., 2007. Electoral competition and agricultural support in OECD countries. Am. J. Polit. Sci. 51, 314–329 Persson, T., Tabellini, G., 2000. Political Economics, Explaining Economic Policy. MIT Press, Cambridge, MA. Persson, T., Tabellini, G., 2002. Do Electoral Cycles Differ across Political Systems? IIES, Stockholm University, Mimeo. Persson, T., Tabellini, G., 2003. The Economic Effect of Constitutions. Cambridge (MA), MIT Press.

Pitlik, H., Wirth, S., 2003. Do crises promote the extent of economic liberalization? An empirical test. Eur. J. Polit. Econ. 19, 565–581. Potrafke, N., 2011. Does government ideology influence budget composition? Empirical evidence from OECD countries. Econ. Gov. 12, 101–134. Roodman, D., 2006. How to Do Xtabond2, An Introduction to Difference and System GMM in Stata. Center for Global Development Working Paper 103. Rozelle, S., Swinnen, J., 2004. Success and failure of reform: insights from the transition of agriculture. J. Econ. Lit. 42, 404–456. Shi, M., Svensson, J., 2006. Political budget cycles, do they differ across countries and why? J. Public Econ. 90, 1367–1389. Swinnen, J., 1994. A positive theory of agricultural protection. Am. J. Agric. Econ. 76, 1–14. Swinnen, J., 1999. The political economy of land reform choices in Central and Eastern Europe. Econ. Transit. 7, 637–664. Swinnen, J., 2002. Major Trends and Developments in the Agribusiness and Agricultural Sectors in CEE and NIS. In: EBRD/FAO Conference on ‘‘Investments in Agribusiness and Agriculture in CEE and the CIS’’, Budapest. Swinnen, J.F.M., 2010a. Agricultural Protection Growth in Europe 1870–1969. In: Anderson, K. (Ed.), The Political Economy of Agricultural Price Distortions. Cambridge University Press, Cambridge, pp. 141–161. Swinnen, J.F.M., 2010b. The political economy of agricultural and food policies: recent contributions, new insights and areas for further research. Appl. Econ. Perspect. Pol. 32, 33–58. Swinnen, J., Gow, J., 1999. Agricultural credit problems and policies during the transition to a market economy in Central and Eastern Europe. Food Policy 24, 21–47. Swinnen, J., Rozelle, S., 2009. Governance structures and resource policy reform, insights from agricultural transition. Annu. Rev. Resour. Econ. 1, 33–54. Thies, C., Porche, S., 2007. The political economy of agricultural protection. J. Polit. 69, 116–127. Ticchi, D., Vindigni, A., 2010. Endogenous constitutions. Econ. J. 120, 1–39. Williamson, J., 1994. In Search of a Manual for Technopols. In: Williamson, J. (Ed.), The Political Economy of Policy Reform. Institute for International Economics, Washington, DC, pp. 11–28. Windmeijer, F., 2005. A finite sample correction for the variance of linear efficient two-step GMM estimators. J. Econom. 126, 25–51. World Bank, 2011. World Development Indicators, CD Rom.