Partisan and Electoral Cycles in Privatisation

Abstract

This study evidences the existence of partisan and electoral cycles in privatisation, by using a sample composed of 22 European countries during the period 1995–2013. Our findings suggest that privatisation reforms tend to be implemented by governments with different ideologies when elections are not immediate; however, results support a greater orientation of right-wing governments towards privatisation reforms in election periods. After the elections, right-wing governments continue adopting privatisation reforms, but governments from ideological traditions tend to oppose such reforms.

Keywords

privatisation political business cycle electoral cycle partisan cycle political ideology elections

Introduction

Following the material approach of Obinger et al. (2016), privatisation may be defined as the sale of the shares of a state-owned enterprise (SOE) to private investors, resulting in a transfer of ownership or voting rights from the public to the private sector.¹ The question of privatisation was initially discussed during the 1920s in France, Switzerland, Germany, Belgium and Italy, aiming to remove the public monopolies on tobacco and railways (Bel, 2011). Partial privatisations were implemented during subsequent decades, until by the late 1980s, privatising SOEs in sectors such as water, energy, telecommunications, or transportation became common.

The British Conservative party of Margaret Thatcher championed privatisation (Clifton et al., 2006). Although such a reform was not included initially in its 1979 election manifesto, it became an important part of the government’s programme to rebalance the public finances and to restructure the national economy towards the private sector (Parker, 2016). Following the UK, other European countries implemented privatisations during the 1980s and 1990s. Traditionally, they have been explained as a way of emulating the UK’s privatisation experience, but several scholars have advocated other reasons, which may be grouped into institutional, political, financial and economic factors (Obinger et al., 2016).² This paper focuses specifically on the political factors, by taking the political business cycle (PBC) approach, which refers to partisan (Hibbs, 1977) and electoral (Nordhaus, 1975) issues.

Political ideology has been one of the most studied factors in the privatisation literature (e.g. see Belke et al., 2007; Bortolotti et al., 2001, 2003; Bortolotti and Pinotti, 2003; Obinger et al., 2014; Opper, 2004; Roberts and Saeed, 2012; Schmitt, 2011, 2013; Schneider and Häge, 2008; Zohlnhöfer et al., 2008). In general, previous studies have evidenced a greater orientation of right-wing governments towards privatisation.

However, the behaviour of politicians is more complex, and it may be influenced by their desire to win the next election. As with several other policies, politicians will be in a hurry to implement the reforms if they expect to obtain some benefit from them (e.g. increasing the probability of winning the election), or they may delay the reforms if negative consequences are expected. For instance, Clifton et al. (2011) indicate that the privatisation of British Telecom was delayed until after the 1983 election for political reasons. Soroka and Wlezien (2010) noted that political parties adjust their policy positions to the electorate’s preferences. Therefore, privatisation policies might be used for opportunistic as well as ideological purposes and the electoral cycle considerations may affect how privatisation proceeds, although the consequences would show up after a significant time lag (Opper, 2004).

Accordingly, we expect the existence of both partisan and electoral cycles in privatisation; privatisation reforms may be explained by ideological motivations, by the proximity of the next election, or by both reasons together. Using a sample composed of 22 European countries during the period 1995–2013, our findings suggest that right-wing governments tend to privatise SOEs to a greater extent than governments with other ideologies (partisan cycle), and that this becomes especially relevant when elections are coming (electoral cycle).

This paper contributes to two lines of research: on the one hand, it contributes to literature on privatisation, showing that political ideology is a determinant of privatisation reforms in Europe, but that they are also affected by the electoral moment. On the other hand, this study contributes to the PBC literature by showing the existence of partisan and electoral cycles in privatisation reforms. Traditionally, PBCs have been studied in terms of macroeconomic variables, such as inflation or unemployment (Hibbs, 1977; Nordhaus, 1975) and budgetary variables, such as expenditures, revenues, fiscal surplus and debt (Philips, 2016). Our findings add new insights in terms of privatisation reforms.

The remainder of the paper is structured as follows: the second section reviews the PBC literature. After that, we propose a hypothesis on the propensity of right-wing governments towards privatisation reforms, taking into account the electoral cycle. The fourth section describes the methodology in detail, and the fifth discusses the descriptive and exploratory analyses. The last section offers conclusions and suggestions for future research.

Political Business Cycles: Background

The term ‘political business cycles’ refers to fluctuations in macroeconomic variables, such as unemployment or inflation, while fluctuations in some components of the budget (e.g. spending, revenues, borrowing, deficits, or debt) are termed as ‘political budget cycles’. In both cases, there are two basic types of models, namely ‘opportunistic’ and ‘partisan’, depending on the reason for why the cycles are induced.

On the one hand, ‘opportunistic’ models are based on the public choice theory (Downs, 1957) that considers politicians as opportunistic agents acting in their own interest (to gain power or to retain it). Accordingly, they try to create favourable conditions in the pre-election period to influence the opinions of voters, thereby increasing their popularity and the probability of being re-elected.

Nordhaus (1975) proposed the traditional ‘opportunistic’ model. However, this model did not take into account the rational expectations of voters, who could be tricked election after election. Rogoff and Sibert (1988) refined the model proposed by Nordhaus (1975) by considering that voters could not be tricked systematically, and they explained ‘opportunistic’ cycles with a temporary information difference between incumbents and voters regarding the competence³ of the government. Voters do not know the current level of competence, and they must infer it. In such a situation, the government may manage macroeconomic variables (business cycles) or budget components (budget cycles) to signal competence in the pre-electoral period.

On the other hand, ‘partisan’ models consider that governments are led by ideological motivations, because each party represents the interests of different segments of the electorate. Those segments are usually placed on the left or right, according to progressive or conservative ideology. In general, lower-income groups tend to support an active state that uses public finances to equalise the outcomes of the market, and higher-income groups tend to support a limited size of the state and a minimum use of public finance, being interested in reducing taxes. Thus, differences between ideologies may lead governments to orient policies, leading to partisan cycles.

Hibbs (1977) developed the first ‘partisan’ model. But, similar to Nordhaus’s model, it assumed irrational expectations. Later, Alesina (1987), who explained fluctuations through partisan or ideological differences in combination with uncertainty about election outcomes, updated it with rational expectations. From the pioneering works of Nordhaus (1975) and Hibbs (1977), an extensive literature has evidenced the existence of opportunistic and partisan cycles in macroeconomic and budgetary variables. In addition, a wave of literature has introduced partisan features in opportunistic cycles, making the two models compatible. That wave was initiated by Frey and Schneider (1978a, 1978b) who stated that governments behave opportunistically when they believe that their chances of winning an election are small, and they behave according to their ideology when the chances of winning are high. These authors found empirical evidence in the UK (Frey and Schneider, 1978a), the US (Frey and Schneider, 1978b), and Germany (Frey and Schneider, 1979).

Since the initial works by Nordhaus (1975) and Hibbs (1977), a vast number of studies have empirically checked opportunistic and partisan cycles (see Dubois, 2016). Although it is impossible to cite all the hundreds of studies published until now, most have tended to focus on four categories of variables to measure the cycles: expenditures, revenues, fiscal surplus and debt (see Philips, 2016). This study contributes to previous literature on PBCs by using a new variable, namely privatisation. We ask if privatisation reforms are induced by the proximity of election, by partisan motivations, or by both together.

The PBC in Privatisation: Research Hypothesis

Traditionally, political ideology has been considered to be a factor that is related to privatisation policies. Since the initial theoretical study by Bienen and Waterbury (1989), which supported ideological rationales behind privatisation reforms, several scholars have analysed the propensity to privatise SOEs by governments of different ideologies in different contexts. In general, most scholars (e.g. Belke et al., 2007; Bortolotti et al., 2001, 2003; Bortolotti and Pinotti, 2003; Obinger et al., 2014; Schmitt, 2013; Schneider and Häge, 2008; Zohlnhöfer et al., 2008) have noted that right-wing governments have a greater predisposition towards privatisation than do leftist governments.

The traditional approach to explain this orientation is based on Hibbs’ (1977) argument. Governments primarily enact policies and reforms that are consistent with the preferences of their core constituencies: left-wing parties tend to represent the interests of the lower-income groups, who generally support an active State; while right-wing parties tend to represent the interests of the higher-income groups, who usually support limited intervention of the State. In such a situation, each party needs the support of the middle class to win the election. For that, right-wing parties may allocate shares of privatised companies to the middle class, who will be averse to the redistribution policies of the left, and will then support the right in the next election (Biais and Perotti, 2002). The rationale is the creation of a middle class of small capitalists who favour market-oriented policies, and thus right-wing parties in the polls (Vickers and Yarrow, 1988).

However, Hibbs’ argument has been criticised as it does not fit well with the current political landscape, which is characterised by large electoral volatility; voters may not have different policy preferences (see Häusermann et al., 2013). For instance, the number of working-class voters has decreased substantially, so left-wing parties have sought to attract other groups of voters, such as socio-cultural professionals and civil servants who are well-educated people, working in the public sector, and with relatively high incomes. These people might have economic policy preferences that are quite distinct from those of the classic working-class electorate of left parties.

However, Engler and Zohlnhöfer (2018) found that the preferences of the (new) middle class differ from the working class only for subsidy spending, but not for regulation and privatisation. Their findings suggest that while the voter preferences−party policy nexus exist for subsidies, this is not the case for regulation and privatisation. So, partisan effects regarding privatisation are not conditional upon the composition of left parties’ electorates. This means that class-based politics may not be the basis of privatisation preferences, since voters have other preferences over and above those ‘dictated’ by their class.

Although several studies support the greater predisposition towards privatisation of right-wing governments, others do not totally support this ideological orientation – see Potrafke (2017) and Zohlnhöfer et al. (2018). We aim to explain this heterogeneity of results regarding the ‘partisan’ model, by considering an additional other essential factor, the proximity of an election (‘electoral’ model).

Baber and Sen (1986) observed that parties of all ideologies address budgetary discipline policies in the years immediately following their election, because voters have a negative opinion about such unpopular reforms, which may negatively affect their voting decisions. Privatisation could be seen as a debated reform because of its controversial effects on some socio-economic factors, such as welfare, equity, employment and corruption (Birdsall and Nellis, 2003; Cuadrado-Ballesteros and Peña-Miguel, 2018; Kikeri and Nellis, 2004; Peña-Miguel and Cuadrado-Ballesteros, 2019). Accordingly, considering privatisation as a controversial reform, we could expect privatisations to be implemented after an election, such as was the case of British Telecom after the 1983 elections (Clifton et al., 2011).

Breen and Doyle (2013) noted that governments are cognisant of the damaging political after-effects of privatisation, and so they will reduce privatisation reforms when elections are close. Therefore, we expect that the partisan orientation towards privatisation is affected by the proximity of the election, which may explain the heterogeneity of results regarding the ideological orientation. Accordingly, we propose the following hypothesis:

Hypothesis. The greater orientation to privatisation of right-wing governments (partisan cycle) is affected by the electoral cycle.

Methodology

Sample

To test the effect of political factors on privatisation reforms, we selected a sample of 22 European countries: Austria, Belgium, the Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, the Netherlands, Poland, Portugal, Slovakia, Slovenia, Spain, Sweden, and the UK. The European context is worth studying due to the importance of privatisation there. Most of the countries in the Organisation for Economic Co-operation and Development (OECD’s) privatisation top 10 are European: France, Italy, Germany, the Netherlands, the UK, Finland and Sweden are among the top 10 if privatisation is measured by the number of transactions, whereas Slovakia, the Czech Republic, Finland, Hungary, Greece, Portugal, France and Poland are among the leading nations when privatisation is measured relative to gross domestic product (GDP) (OECD, 2009).

The period of analysis was 1995–2013; thus, a panel data set consisting of 418 observations was used in the analyses. The period has been selected on the basis of the availability of data, especially on privatisation, which were obtained from the Privatisation Barometer website. This is a project launched by Fondazione Eni Enrico Mattei (FEEM),⁴ a non-profit and non-partisan institution for the study of governance. Privatisation Barometer offers information about privatisations (total and partial) in Europe until 2013. This is the reason the period of analysis ends in 2013.

Data on elections and political ideology were obtained from the Comparative Political Data Set 1960−2016 (Armingeon et al., 2018). The rest of the variables, regarding the budgetary and socio-economic factors that we used as control variables, were obtained from the Eurostat, OECD and World Bank databases.

Variables

First, we needed some indicators to represent privatisation reforms. Accordingly, we created two variables: Deals refers to the number of privatisation transactions (both partial and total) conducted in a country by year, and the Proceeds variable represents the total revenue (in current US$) from privatisation deals as a proportion of GDP (in current US$) (Bortolotti et al., 2001, 2003; Zohlnhöfer et al., 2008). The first variable (Deals) represents privatisation policy progress, whereas the second one (Proceeds) measures the economic impact of such reforms (Bortolotti et al., 2003). It is important to include the two indicators because, on the one hand, considering just the number of transactions underestimates the economic effect of privatisation, but on the other hand, considering only the revenue overestimates the impact of privatisation when only a few large SOEs are involved (Bortolotti et al., 2001).

Second, we needed some variables representing the electoral cycle and the political ideology of the government. Concretely, we count the number of years left in the current term until the next election, creating the variable Elections; this takes the value of 0 in the election year, 1 in the first year before an election, 2 in the second year prior to an election, and so on. The political ideology of the government is represented by the variable Right, which refers to the government composition by the relative power of right-wing parties (liberal and conservative) in government. It is calculated as

R i g h t = \frac{\begin{array}{l} Parliamentary seat share of right \\ - wing parties in government \end{array}}{\begin{array}{l} Parliamentary seat share of \\ all parties in government \end{array}} * \frac{Number of days in office}{365}

Third, the analyses control for several factors that explain privatisation reforms: Debt is the government consolidated gross debt (% of GDP), and it is expected to be positively related to privatisation reforms (Bortolotti et al., 2003; Jeronimo et al., 2000); FDI represents the foreign direct investment, net inflows (% of GDP), as a proxy of the openness of the economy, which may facilitate the sale of SOEs (Belloc et al., 2014); the Market_cap variable is the market capitalisation of listed domestic companies (% of GDP), which represents the development of the stock market, and it is expected that it facilitates privatisation reforms (Bortolotti et al., 2001, 2003); Trade_union is the number of union members divided by the total number of employees (Jahn, 2016); and the income level is represented by GDPpc. the GDP per capita (Bortolotti et al., 2003).

Model and Technique

The proposed hypothesis will be empirically tested by estimating the following models:

\begin{array}{l} P r i v a t i s a t i o n_{i t} = β_{1} E l e c t i o n s_{i t} + β_{2} R i g h t_{i t} + β_{3} D e b t_{i t} + β_{4} F D I_{i t} + \\ β_{5} M a r k e t_c a p_{i t} + β_{6} T r a d e_u n i o n_{i t} + β_{7} G D P p c_{i t} + \sum_{j = 8}^{26} β_{j} Y e a r_{t} + η_{i} + ν_{i t} \end{array}

(1)

\begin{array}{l} P r i v a t i s a t i o n_{i t} = β_{1} E l e c t i o n s_{i t} + β_{2} R i g h t_{i t} + α E l e c t i o n s * R i g h t_{i t} + β_{3} D e b t_{i t} + \\ β_{4} F D I_{i t} + β_{5} M a r k e t_c a p_{i t} + β_{6} T r a d e_u n i o n_{i t} + β_{7} G D P p c_{i t} + \sum_{j = 8}^{26} β_{j} Y e a r_{t} + η_{i} + ν_{i t} \end{array}

(2)

where subindexes i and t refer to each country and year, respectively; β represents parameters to be estimated; Privatisation refers to indicators of privatisation reforms, namely Deals and Proceeds; the remaining variables have been previously defined, and Elections*Right is the interaction term between those two variables, η_i refers to characteristics of countries that differ among them but are invariant over time (i.e. unobservable heterogeneity), and ν_it is the classic disturbance term.

The models are estimated using econometric techniques for panel data. The fixed or random effects (FE or RE) are the traditional estimators for panel data models, but both require homoscedasticity and no serial correlated errors. Thus, first, we tested these conditions using the Breusch–Pagan test and the Wooldridge test, respectively. The p-values obtained were lower than 0.05, so we rejected the null hypotheses of homoscedastic errors and no serially correlated errors. Therefore, neither FE nor RE estimators are appropriate in this case, since the two conditions are violated.

Second, endogeneity problems also appeared in the two models due to the three reasons noted by Wooldridge (2010): (1) the use of proxy variables to represent concepts that are difficult to represent, such as most of the political factors, because they are not directly observed; (2) the possibility that the results could additionally be controlled by other variables (e.g. education level, income level, unemployment, level of corruption, globalisation, etc.), which have been omitted due to multicollinearity problems with other control variables; and (3) the existence of reverse causality between the dependent variable (privatisation) and most of the control variables, such as Debt, GDP, FDI, or market capitalisation.

Accordingly, endogeneity should be addressed using instrumental variable (IV) methods. In the presence of heteroscedasticity (as is the case), the conventional IV estimator is consistent but inefficient (Baum et al., 2003). Therefore, it is necessary to use another IV method, such as the dynamic panel estimator (Arellano and Bond, 1991), which overcomes such a limitation. Concretely, we used the two-step system estimator of Arellano and Bover (1995), which augments the traditional estimator, namely the difference estimator (Arellano and Bond, 1991). The latter is consistent when the sample size is large, but it has poor small-sample properties, and the system estimator is able to solve this problem (Arellano and Bover, 1995; Blundell and Bond, 1998).

The system estimator uses the lagged values of endogenous and predetermined variables as instruments to correct endogeneity. It has been demonstrated that these instruments are uncorrelated with the error term (Arellano and Bond, 1991) and that they usually contain better information on the current value of the variable than outside instruments. However, this approach may lead to a proliferation of instruments, and the results could be biased. Accordingly, the instrument validity is tested by (1) the Arellano–Bond test for AR(2) in the first differences, under the null hypothesis of no serial correlation between the error terms and (2) the Hansen test of overidentification restrictions, under the null hypothesis that the overidentifying restrictions are valid. The results of these tests are shown at the bottom of Tables 3 and 4.

Results

Descriptive Statistics

Table 1 shows the descriptive statistics of the variables described previously.

Table 1.

Descriptive Statistics.

Variable	Observation	Mean	Standard deviation	Minimum	Maximum
Deals	418	4.3469	5.4131	0	55
Proceeds	418	0.4450	0.7251	0	5.16
Elections	416	1.8582	1.2521	0	4
Right	418	40.2914	37.4716	0	100
Debt	413	55.9196	31.3665	3.7	177.4
FDI	410	5.4182	8.3754	–15.99	87.44
Market_cap	302	51.7604	38.8304	1.19	258.38
Trade_union	418	32.0858	19.9467	6.3589	86.6213
GDPpc	418	52,962.87	118,325.60	2168.80	808,886.60

FDI: foreign direct investment.

The mean value of Deals suggests that there were around four or five privatisation transactions a year in each of the sample countries during the period of analysis, on average. The mean value of Proceeds indicates that the value of transactions is 0.44% of GDP, on average. Nevertheless, there are large differences: the maximum value of Deals is 55, which was obtained by Poland in 2010, while the maximum value of Proceeds is 5.16, obtained by Portugal in 1997 when the government sold, among others, 30% of Electricidade de Portugal (EDP) for US$2,033.4 million. Thus, although Poland is the country with the largest number of privatisation transactions, Portugal excels with revenues obtained from privatisation reforms. These differences can also be seen in Figure 1, in which the mean values of Deals and Proceeds are represented by country.

Figure 1.

Privatisations by Country.

Furthermore, we can see the evolution of privatisation indicators in Figure 2. In general, there is no clear trend: between 1995 and 1998, although the number of transactions reduced, the economic relevance of such sales increased, on average. From that moment, there is a reduction both in the number and monetary value of privatisation transactions until 2010, although the Proceeds variable increases again between 2003 and 2008. From 2010, there was a tendency to privatise in Europe.

Figure 2.

Evolution of Privatisation (1995–2013).

Returning to Table 1, the mean value of Right suggests 49.29% of the total parliamentary seat share of all governing parties is held by rightists. Regarding the control variables, the government consolidated gross debt (Debt) is about 56% of GDP, on average, although such value increases to 177.4% in Greece in 2013, and it is 3.7% in Estonia in 2007. The net inflow of investment (FDI) is about 5.4% of GDP on average, ranging from -16% in Hungary in 2010 to 87.44% in the Netherlands in 2007. The mean value of the Market_cap variable indicates that the market capitalisation of listed domestic companies is 51.76%, ranging from 258.4% in Finland (1999) to 1.19% in Austria (1998). The mean value of Trade_union indicates 32.08% of employees are union members, although the percentage falls to about 6% in Estonia and it increases up to more than 80% in Sweden.

Table 2 shows the bivariate correlations between variables entered into the models. Correlations between independent/control variables are not very high and most of them are not statistically relevant. In descriptive terms, most of the correlation coefficients were less than 0.4, which is an accepted threshold for multicollinearity problems (Wooldridge, 2010).

Table 2.

Bivariate Correlations.

	Deals	Proceeds	Elections	Right	Debt	FDI	Market_cap	Trade_union	GDPpc
Deals	1
Proceeds	0.2632***	1
Elections	–0.0182	–0.012	1
Right	0.0798	0.0405	–0.0309	1
Debt	0.0927^†	0.1114*	0.0456	0.2365***	1
FDI	–0.1247*	0.0541	0.0467	0.0775	–0.0637	1
Market_cap	–0.0243	0.0537	–0.0222	0.0411	–0.0252	0.1972***	1
Trade_union	–0.0852^†	–0.0876^†	–0.0717	–0.0546	0.1072*	–0.0291	0.218***	1
GDP	–0.153**	–0.1583**	0.0922^†	–0.0719	–0.0819	–0.0794	–0.1141*	0.2304***	1

FDI: foreign direct investment; GDP: gross domestic product.

†

Significant at 10% level; *significant at 5% level; **significant at 1% level and ***significant at 0.1% level.

Empirical Results

Empirical results for Models 1 and 2 are shown in Table 3: panel A shows the results for the dependent variable Deals, and panel B shows the results for Proceeds. In addition, at the bottom of each equation we can see the results for the Arellano and Bond test for AR(2) and the Hansen test; the p-values indicate that we cannot reject the null hypotheses of: (1) no serial correlation between the error terms and (2) the overidentifying restrictions are valid. Thus, this confirms the validity of instruments.

Table 3.

PBC in Privatisation Reforms.

Panel A. Effect on deals variable
	Equation (1)		Equation (2)
	Coefficient.	Standard error	Coefficient	Standard error
Elections	–0.0823^†	0.0393	0.5346***	0.0774
Right	0.0110*	0.0044	0.0375**	0.0123
Elections × Right			–0.0144***	0.0034
Debt	0.0097*	0.0041	0.0209**	0.0062
FDI	0.1186*	0.0232	0.1482**	0.0433
Market_cap	0.0098*	0.0046	0.0016	0.0070
Trade_union	–0.0209^†	0.0108	–0.0283	0.0357
GDPpc	–0.0071***	0.0011	–0.0093**	0.0032
Arellano-Bond test for AR(2)	Pr > z = 0.352		Pr > z = 0.533
Hansen test	Pr > z = 1.000		Pr > z = 1.000
Panel B. Effect on proceeds variables
	Equation (1)		Equation (2)
	Coefficient.	Standard error	Coefficient	Stdandard error
Elections	–0.0050*	0.0021	0.3339***	0.0709
Right	0.0531*	0.0231	0.0144***	0.0030
Elections × Right			–0.0064***	0.0013
Debt	0.0036	0.0033	–0.0024	0.0027
FDI	0.0022	0.0088	–0.0047	0.0151
Market_cap	–0.0021	0.0015	0.0038*	0.0014
Trade_union	–0.0157*	0.0058	–0.0061	0.0067
GDPpc	–0.0006**	0.0001	–0.0026**	0.0009
Arellano-Bond test for AR(2)	Pr > z = 0.559		Pr > z = 0.756
Hansen test	Pr > z = 1.000		Pr > z = 1.000

FDI: foreign direct investment.

Dependent variables: Deals refers to the number of privatisation transactions (both partial and total); Proceeds variable represents the total revenues from privatisation deals as a proportion of GDP. All regressions include year fixed effects. AR(2) is the second-order serial correlation. In our models it analyses whether privatisations in period ‘t’ are affected or not by privatisations in period ‘t-2’.

†

Significant at 10% level; *Significant at 5% level; **Significant at 1% level and ***Significant at 0.1% level.

The Elections variable impacts negatively on Deals, such as we can see in equation (1) of panel A, and it is statistically relevant at 90% confidence level. This result indicates that the number of privatisation transactions increases as the number of years until the next election is reduced, that is, privatisation reforms tend to be carried out when elections are close. This finding evidences the existence of electoral cycles in privatisation reforms.

Political ideology is also a determinant of privatisations, since the Right variable impacts positively on Deals and is statistically relevant at 95%. This means that the higher the proportion of right-wing parties in the government, the greater the number of privatisations carried out.

Despite these individual effects, opportunistic and partisan cycles may be interrelated. Accordingly, we estimate Model 2. Right again impacts positively on Deals, so the empirical results note that privatisation transactions increase as the power position of right-wing parties in government increases. Furthermore, this effect increases as elections get closer, since the interaction term Elections*Right has a negative coefficient. These findings suggest that right-wing governments tend to implement more privatisation reforms than other ideologies, but such implementation is especially relevant when the elections are close to being held. Thus, privatisations may be seen as an electoral tool used by right-wing government in the pre-election period. This is in accord with the hypothesis proposed initially, making compatible partisan and electoral cycles in terms of privatisation reforms.

Regarding other ideologies, if Right is reduced (i.e. the relative power position of right-wing parties in government is reduced), the interaction term will also be reduced; and then, ‘other ideologies’ are represented by the coefficient of Elections, which is positive and relevant at 99.9% confidence level. This means that governments with other ideologies may privatise SOEs when the number of years left until the next election increases; while right-wing governments are always likely to privatise, but that they privatise more as elections get closer.

These results are similar to those obtained for Proceeds in panel B. In equation (1), the Elections variable negatively impacts the dependent variable and Right has a positive coefficient; the two are statistically relevant at 95%. In equation (2), Right continues impacting positively on the privatisation indicator (Proceeds), and the interaction term Election*Right has a negative coefficient. These findings indicate that right-wing governments tend to privatise SOEs to a greater extent than centrist and leftist governments, which is especially relevant when elections are coming.

Regarding control variables, the level debt (Debt) and foreign direct investment (FDI) are positively related with Deals. Market capitalisation (Market_cap) is, in general, positively associated with privatisations, while trade union density (Trade_union) affects negatively. GDP per capita is negatively related with privatisation reforms (Deals and Proceeds).

Robustness Checking

Two robustness analyses have been carried out by changing the core variables, Right and Elections. The results are included in Table 4.

Table 4.

Robustness Checking.

Panel A. Changing ideology variable
	Equation (1): Deals		Equation (2): Proceeds
	Coefficient	Standard error	Coefficient	Standard error
Elections	0.3493*	0.1214	0.2647*	0.1001
Right_support	0.0555**	0.0156	0.0202*	0.0095
Elections × Right_support	–0.0176**	0.0057	–0.0094*	0.0037
Debt	0.0190**	0.0062	–0.0012	0.0027
FDI	0.1471**	0.0440	0.0128	0.0128
Market_cap	0.0047	0.0057	0.0059***	0.0013
Trade_union	–0.0234	0.0369	0.0002	0.0092
GDPpc	–0.0085**	0.0025	–0.0019^†	0.0010
Arellano-Bond test for AR(2)	Pr > z = 0.518		Pr > z = 0.766
Hansen test	Pr > z = 1.000		Pr > z = 1.000
Panel B. Changing election variable
	Equation (1): Deals		Equation (2): Proceeds
	Coefficient	Standard error	Coefficient	Standard error
D_Elections	–0.0130***	0.0011	–0.0026*	0.0010
Right_support	0.0202*	0.0087	0.0091*	0.0043
D_Elections × Right_support	0.0161***	0.0036	0.0098*	0.0038
Debt	0.0022	0.0157	0.0008	0.0036
FDI	0.0372^†	0.0196	0.0040	0.0078
Market_cap	0.0170**	0.0055	0.0049*	0.0019
Trade_union	–0.2885***	0.0265	0.0111	0.0157
GDPpc	–0.0075***	0.0008	–0.0018^†	0.0009
Arellano-Bond test for AR(2)	Pr > z = 0.185		Pr > z = 0.516
Hansen test	Pr > z = 1.000		Pr > z = 1.000

FDI: foreign direct investment.

Dependent variable equation (1): Deals refers to the number of privatisation transactions (both partial and total). Dependent variable equation (2): Proceeds variable represents the total revenues from privatisation deals as a proportion of GDP. All regressions include year fixed effects. AR(2) is the second-order serial correlation. In our models it analyses whether privatisations in period ‘t’ are affected or not by privatisations in period ‘t-2’.

†

Significant at 10% level; *significant at 5% level; **significant at 1% level and ***significant at 0.1% level.

First, the partisan variable has been replaced by Right_support, which refers to the government support in Parliament (instead of the government composition represented by Right). Concretely, it is measured as:

\begin{array}{l} R i g h t_s u p p o r t = Parliamentary seat share of right - wing parties in government \\ * \frac{Number of days in office}{365} \end{array}

While Right takes into account the percentage of seats of the government that are held by right-wing parties, Right_support takes into account the seats of Parliament that are occupied by the rightists that are in government. So this indicator takes into account the relevance of right-wing ideologies against other ideologies in Parliament. The results of Model (2), which is more complete than Model (1), are shown in panel A of Table 4, and they are in accordance with those obtained previously.

Second, the variable Elections has been replaced by D_Elections, which is a dummy variable that takes the value 1 if there was an election in that year and also it takes the value 1 in the year before that election. The reason for changing this variable is that early elections cannot be taken into account appropriately with the previous variable Elections, which can jump, for example, from Elections = 2 to Elections = 0. Meanwhile, D_Elections takes only two values, 1 for the pre-election period, and 0 at other times. So, we coded D_Elections = 1 in the year when elections were held, and then we also assigned the value 1 to the first-order lag.

The empirical results of model (2) are shown in panel B of Table 4. We can see that Right_support has a positive impact on privatisation reforms (both Deals and Proceeds), which is in accord with the partisan effect that was found previously. Furthermore, this effect is increased in the pre-election period, since the interaction term is positive and statistically relevant in both equations (Deals and Proceeds). These findings also accord with those obtained previously. However, as right-wing government support is reduced, the privatisation reforms in the electoral period are also reduced, since the coefficient of D_Elections is negative and relevant in the both equations.

Figure 3 shows the level of privatisations⁵ implemented in the two electoral moments, that is, in the pre-election period when elections are close to being held (D_Elections = 1), and in the years after the elections (D_Elections = 0). As we can see, the orientation towards privatisation increases as the government tend to the right-wing ideology, and that partisan effect is even larger in the pre-election period. These findings are shown for Deals and also for Proceeds.⁶

Figure 3.

Marginal Effects on Privatisation.

Discussion and Conclusion

Following the empirical analysis of a sample of 22 European countries over the period 1995–2013, we can state that privatisation reforms are affected by both partisan and electoral cycles. Our results support the proposed hypothesis, suggesting that the greater orientation towards privatisation by right-wing governments is affected by the cyclicality of elections. More concretely, our findings indicate that privatisation reforms tend to be carried out by right-wing governments, such as other scholars have noted previously (e.g. Belke et al., 2007; Bortolotti et al., 2001, 2003; Bortolotti and Pinotti, 2003; Schmitt, 2013; Schneider and Häge, 2008; Zohlnhöfer et al., 2008), and especially if the next election is close, that is, in the election year, and the years preceding and following the election. Then, privatisation seems to be used as an electoral tool by conservative governments: in allocating shares of privatised companies to the electorate, the rightists try to attract their vote, to increase their probability of re-election.

This study makes relevant contributions to different lines of research. On the one hand, it contributes to the PBC literature, by showing a fluctuation in privatisation policies, which is induced by the cyclicality of elections and by partisan features. Previous studies on the PBC are focused on expenditure, revenue or debt; this paper contributes by showing partisan and electoral cycles in privatisations. On the other hand, it adds evidence to the political economy of privatisation literature by corroborating the relevance of the political framework in explaining privatisation reforms (Obinger et al., 2016). Our findings contribute to conclusions obtained by Clifton et al. (2006), who placed European privatisation in the context of economic and political integration. Our empirical results suggest that partisan and electoral issues matter, and they should be taken into account (additionally to other factors) in explaining privatisations in Europe.

Despite these contributions, this study is not free of limitations. One of them is that our data are at the country−year level and sectorial issues are not easy to control. In addition, the period of analysis ends in 2013, due to the availability of data; so this study does not take into account the most recent privatisations in Europe, promoted by the Troika, as a way to recover the financial situation in some countries (Clifton et al., 2018). It would be interesting to include this wave of privatisation in future studies, by using other sources of data that include more recent reforms.

It would also be interesting to consider the replication of this study covering other regions, such as Latin America or Asia, which allows the controlling of contextual factors, such as the governance quality, the level of democracy, the economic development, and so on. Finally, future research could consider the partisan preferences and the new challenges in the globalised world, such as the sustainability of public pensions, the consequences of Brexit, and the multi-collateral effects of migration policy.

Footnotes

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work has been supported by the Consolidated Research Group EJ/GV: IT 897-16.

ORCID iDs

Noemí Peña-Miguel

Beatriz Cuadrado-Ballesteros

Notes

Author Biographies

Noemí Peña-Miguel is Assistant Professor of Accounting at the University of the Basque Country where she earned her PhD in Business. She conducts her teaching and research activity on welfare, social security, and finance and accounting.

Beatriz Cuadrado-Ballesteros is Lecturer in Accounting at the University of Salamanca, where she earned her PhD in Business. Her research interests are public management, political science and public sector reforms.

References

Alesina

(1987) Macroeconomic Policy in a Two-Party System as a Repeated Game. The Quarterly Journal of Economics 102: 651–678.

Arellano

Bond

(1991) Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations. The Review of Economic Studies 58: 277–297.

Arellano

Bover

(1995) Another Look at the Instrumental Variables Estimation of Error Components Models. Journal of Econometrics 68: 29–51.

Armingeon

Wenger

Wiedemeier

et al. (2018) Comparative Political Data Set 1960–2016. Bern: Institute of Political Science, University of Bern.

Baber

Sen

(1986) The Political Process and the Use of Debt Financing by State Governments. Public Choice 48 (3): 201–215.

Baum

Schaffer

Stillman

(2003) Instrumental Variables and GMM: Estimation and Testing. The Stata Journal 3 (1): 1–31.

Bel

(2011) The First Privatisation: Selling SOEs and Privatising Public Monopolies in Fascist Italy (1922–1925). Cambridge Journal of Economics 35 (5): 937–956.

Belke

Baumgärtner

Schneider

et al. (2007) The Different Extent of Privatization Proceeds in OECD Countries: A Preliminary Explanation Using a Public-Choice Approach. Finanzarchiv: Public Finance Analysis 63 (2): 211–243.

Belloc

Nicita

Sepe

(2014) Disentangling Liberalization and Privatization Policies: Is There a Political Trade-Off? Journal of Comparative Economics 42 (4): 1033–1051.

10.

Biais

Perotti

(2002) Machiavellian Privatisation. The American Economic Review 92 (1): 240–258.

11.

Bienen

Waterbury

(1989) The Political Economy of Privatization in Developing Countries. World Development 17 (5): 617–632.

12.

Birdsall

Nellis

(2003) Winners and Losers: Assessing the Distributional Impact of Privatisation. World Development 31 (10): 1617–1633.

13.

Blundell

Bond

(1998) Initial Conditions and Moment Restrictions in Dynamic Panel Data Models. Journal of Econometrics 87 (1): 115–143.

14.

Bortolotti

Pinotti

(2003) The Political Economy of Privatization. Fondazione Eni Enrico Mattei (FEEM) Working Paper no. 45.2003, June. Milano: Fondazione Eni Enrico Mattei.

15.

Bortolotti

Fantini

Siniscalco

(2001) Privatisation: Politics, Institutions, and Financial Markets. Emerging Markets Review 2 (2): 109–137.

16.

Bortolotti

Fantini

Siniscalco

(2003) Privatisation around the World: Evidence from Panel Data. Journal of Public Economics 88 (1–2): 305–332.

17.

Breen

Doyle

(2013) The Determinants of Privatization: A Comparative Analysis of Developing Countries. Journal of Comparative Policy Analysis: Research and Practice 15 (1): 1–20.

18.

Clifton

Comín

Díaz-Fuentes

(2006) Privatizing Public Enterprises in the European Union 1960–2002: Ideological, Pragmatic, Inevitable? Journal of European Public Policy 13 (5): 736–756.

19.

Clifton

Comín

Díaz-Fuentes

(2011) From National Monopoly to Multinational Corporation: How Regulation Shaped the Road towards Telecommunications Internationalisation. Business History 53 (5): 761–781.

20.

Clifton

Diaz-Fuentes

Gómez

(2018) The Crisis as Opportunity? On the Role of the Troika in Constructing the European Consolidation State. Cambridge Journal of Regions, Economy and Society 11 (3): 587–608.

21.

Cuadrado-Ballesteros

Peña-Miguel

(2018) The Socioeconomic Consequences of Privatization: An Empirical Analysis for Europe. Social Indicators Research 139 (1): 163–183.

22.

Downs

(1957) An Economic Theory of Political Action in a Democracy. Journal of Political Economy 65 (2): 135–150.

23.

Dubois

(2016) Political Business Cycles 40 Years after Nordhaus. Public Choice 166 (1–2): 235–259.

24.

Engler

Zohlnhöfer

(2018) Left Parties, Voter Preferences, and Economic Policy-Making in Europe. Journal of European Public Policy. Epub ahead of print 15 November. DOI: 10.1080/13501763.2018.1545796.

25.

Frey

Schneider

(1978a) A Politico-Economic Model of the United Kingdom. The Economic Journal 88: 243–253.

26.

Frey

Schneider

(1978b) An Empirical Study of Politico-Economic Interaction in the United States. The Review of Economics and Statistics 60: 174–183.

27.

Frey

Schneider

(1979) An Econometric Model with an Endogenous Government Sector. Public Choice 34 (1): 29–43.

28.

Gonzalo

Pina

Torres

(2003) Objectives, Techniques and Valuation of State-Owned Companies in Privatization Processes. Public Management Review 5 (2): 177–196.

29.

Häusermann

Picot

Geering

(2013) Review Article: Rethinking Party Politics and the Welfare State – Recent Advances in the Literature. British Journal of Political Science 43 (1): 221–240.

30.

Hibbs

(1977) Political Parties and Macroeconomic Policy. American Political Science Review 71 (4): 1467–1487.

31.

Jahn

(2016) Changing of the Guard: Trends in Corporatist Arrangements in 42 Highly Industrialized Societies from 1960 to 2010. Socio-Economic Review 14 (1): 47–71.

32.

Jeronimo

Pagán

Soydemir

(2000) Privatization and European Economic and Monetary Union. Eastern Economic Journal 26 (3): 321–333.

33.

Kikeri

Nellis

(2004) An Assessment of Privatization. The World Bank Research Observer 19 (1): 87–118.

34.

Nordhaus

(1975) The Political Business Cycle. The Review of Economic Studies 42 (2): 169–190.

35.

Obinger

Schmitt

Traub

(2016) The Political Economy of Privatization in Rich Democracies. Oxford: Oxford University Press, 156 pp.

36.

Obinger

Schmitt

Zohlnhöfer

(2014) Partisan Politics and Privatization in OECD Countries. Comparative Political Studies 47 (9): 1294–1323.

37.

Organisation for Economic Co-operation and Development (OECD) (2009) Privatisation in the 21st Century: Recent Experiences of OECD Countries. Available at: http://www.oecd.org/daf/ca (accessed 10 July 2017).

38.

Opper

(2004) The Political Economy of Privatization: Empirical Evidence from Transition Economies. Kyklos 57 (4): 559–586.

39.

Parker

(2016) Privatization of the United Kingdom’s Public Utilities: The Birth of the Policy 1979–1984. Annals of Public and Cooperative Economics 87 (1): 5–22.

40.

Peña-Miguel

Cuadrado-Ballesteros

(2019) Is Privatization Related to Corruption? An Empirical Analysis of European Countries. Public Management Review 21 (1): 69–95.

41.

Philips

(2016) Seeing the Forest through the Trees: A Meta-Analysis of Political Budget Cycles. Public Choice 168 (3–4): 313–341.

42.

Potrafke

(2017) Partisan Politics: The Empirical Evidence from OECD Panel Studies. Journal of Comparative Economics 45 (4): 712–750.

43.

Roberts

Saeed

(2012) Privatizations around the World: Economic or Political Determinants? Economics & Politics 24 (1): 47–71.

44.

Rogoff

Sibert

(1988) Elections and Macroeconomic Policy Cycles. The Review of Economic Studies 55 (1): 1–16.

45.

Schmitt

(2011) What Drives the Diffusion of Privatisation Policy? Evidence from the Telecommunications Sector. Journal of Public Policy 31 (1): 95–117.

46.

Schmitt

(2013) The Janus Face of Europeanisation: Explaining Cross Sector Differences in Public Utilities. West European Politics 36 (3): 547–563.

47.

Schneider

Häge

(2008) Europeanization and the Retreat of the State. Journal of European Public Policy 15 (1): 1–19.

48.

Soroka

Wlezien

(2010) Degrees of Democracy: Politics, Public Opinion, and Policy. Cambridge: Cambridge University Press.

49.

Vickers

Yarrow

(1988) Privatisation: An Economic Analysis. Cambridge, MA: The MIT Press.

50.

Wooldridge

(2010) Econometric Analysis of Cross Section and Panel Data. Cambridge, MA: The MIT Press.

51.

Zohlnhöfer

Engler

Dümig

(2018) The Retreat of the Interventionist State in Advanced Democracies. British Journal of Political Science 48 (2): 535–562.

52.

Zohlnhöfer

Obinger

Wolf

(2008) Partisan Politics, Globalization, and the Determinants of Privatisation Proceeds in Advanced Democracies (1990–2000). Governance: An International Journal of Policy, Administration, and Institutions 21 (1): 95–112.