Eduard-Alex Ciuhandu
Comparative Politics
Michaelmas Term, 2024
Cambridge Journal of Political Affairs, 5(2), pp. 149–172

https://doi.org/10.5281/zenodo.14354985

1. Introduction

Recent events in the political sphere have shown that there is renewed interest and controversy surrounding the topic of pensions. Instead of solely being a topic in public economics, pensions and pension systems have become an integral part of everyday politics. This paper explores political factors that influence the decision of countries to adopt a mandatory private pension pillar. To showcase the political controversies surrounding pensions, I will first present two recent political developments in France and Germany, incorporating them into the larger debate on the structure of old age security in the 21st century.

The most striking recent event regarding pensions is arguably the pension reform in France and the massive wave of protests that followed (Corbet and Ganley, 2023; France24, 2023). Introduced by President Macron to make the country more economically competitive and prevent the pension system from going into deficit (Corbet, 2023), the controversy surrounding the reforms dominated European and worldwide news. While the president was concerned with economic factors, the protesters demanded that the principles of intergenerational solidarity be maintained, proposing higher taxes on high-income earners as the optimal solution. The debate also reached the academic sphere, with economists taking a stance. One of the most vocal critics of the reform was Thomas Piketty, who argued that the necessary funds could be generated through higher taxes on top earners and accused the French president of being a ‘president of the rich’ (Piketty, 2023).

Another recent development regarding pensions took place in Germany, where the liberal finance minister introduced the idea of a stock pension (Aktienrente) to address the sustainability issues of the public pension system (Serenelli, 2022; Zeit Online, 2022). The justification for this proposal is the challenges faced by Germany in terms of financing public pensions due to demographic changes caused by the ageing population—as life expectancy increases and the birth rate decreases, the pensions of the increasing retiree population have become more dependent on the contributions paid by a decreasing number of economically active individuals (Deutschlandfunk, 2024). Originally proposed as a mandatory private pillar resembling the Swedish model (Schier et al., 2021), the outcome is more similar to the Norwegian model, consisting of a state-funded and independently managed investment fund, with the government planning to allocate ten billion euros yearly to the fund (Deutschlandfunk, 2024). The debate surrounding the German proposal bears similarities to the French case, with the liberals pushing for market-based solutions and the Greens and Left being cautious or outright opposing such an approach.

These two recent examples illustrate a bigger issue in European countries, where the sustainability of the public pension system has become endangered by the ageing phenomenon (Amaglobeli et al., 2019; Corselli-Nordblad and Strandell, 2020). They also show how different political positions can lead to different proposals to solve this problem: for instance, the original proposal of the German liberals brings the idea of a mandatory private pension pillar back to the spotlight after it faded away after the 2008 financial crisis. One of the questions regarding pension reform thus is whether it is a political decision or solely an economic one based on demographics. Therefore, I define my research question as such: what political factors influence a country’s decision to adopt a mandatory private pension pillar?

As pension systems and reforms are often hard to understand due to the focus on technical and complex details, I will first provide a clear definition and an overview of their emergence and evolution. I define mandatory private pillars (MPPs) by three fundamental characteristics: mandatory, private, and individual. Firstly, participation is compulsory for the working population. As such, it differentiates itself from voluntary private pensions, where one may choose whether or not to contribute. The second characteristic means that an MPP is wholly managed by the private sector, clearly separating such a pillar from the classical Bismarckian publicly managed pensions. Thirdly, I restrict my understanding of MPPs to individual-oriented and not occupational pillars. MPPs are thus mostly defined-contribution systems, where the individual pension depends on the accumulated monetary contributions and the interest earned. MPP adoptions, unlike other forms of pension reforms, are structural, as they complement or replace the defined-benefits (DB) pay-as-you-go (PAYG) pillars (Kohli and Arza, 2011).

Such a pension system is a relatively new concept, but locating the exact roots of the idea can be difficult. While some scholars point to Switzerland or Singapore (Vittas, 1993), it was the Chilean pension reform that brought the MPP concept to public attention. To solve its major pension problems, Chile introduced mandatory pension accounts for formal sector workers that were managed by the private sector through diverse portfolios of stocks and bonds (Barr et al., 2008). This policy, implemented in 1981 as part of a broader reform to liberalise Chile’s economy, was mostly seen as a success and is credited with having averted higher fiscal deficits (Cerda, 2008). The model’s popularity spread quickly through Latin America, with other countries adopting similar models throughout the next decade (Edwards, 1998). However, the MPP model most commonly known today differs from the Chilean model. Developed in 1994 when the World Bank endorsed the creation of multi-pillar pension systems, the original multi-pillar model consists of three pension pillars:

1. A mandatory public pillar with a redistributive function. The scope of this pillar is to combat poverty among the elderly but should not be expanded beyond what is necessary to reduce the tax burden.
2. A mandatory private pillar that is publicly regulated but privately managed. Its primary role is individual savings based on contributions, although the pillar should also promote capital development and boost economic growth.
3. A voluntary pillar for additional protection for individuals who want a higher pension after retirement.

While the World Bank’s conceptualisation has changed over time into a five-pillar structure with an additional zero and fourth pillar (Holzmann et al., 2008), the core model remains the same: the Bismarckian DB PAYG system (first pillar) is to be complemented by an MPP.

In Europe, this pension pillarisation grew in popularity mostly in Central and Eastern Europe (CEE) at the end of the twentieth century until the 2008 crisis. The first countries to adopt it were Poland and Hungary, quickly followed by other neighbouring countries (Mladen, 2012). The last CEE country to implement an MPP was Romania in 2008, although the popularity of MPPs soon dissipated with the onset of the global financial crisis. In the aftermath of the crisis, most countries reduced the contribution rate to the MPP (Bielawska et al., 2017), and Hungary even repealed the pillar altogether (Simonovits, 2012).

Subsequently, I will systematically review the existing literature for possible determinants of such reforms (section 2). I explore in-depth the role of political ideology and the involvement of transnational actors (section 3). In this context, I hypothesise that both a right-wing government and a World Bank pension project increase the chances of MPP adoption. Empirically, I analyse the current EU countries from 1980 to 2020 (section 4) and the particular case of Romania, also comparing it to World Bank projects for pension reform in six other countries (section 5). Finally, after discussing the results of my mixed methods approach (section 6), I draw conclusions and highlight potential avenues for further research (section 7). My main findings are that both the governing party’s ideology and the World Bank’s involvement are significant factors influencing the adoption and implementation of an MPP, while the ‘classical’ demographic explanations do not yield significant results.

2. Literature Review – (Political) Determinants For Pension Reform

As the adoption of MPPs in countries that previously had PAYG DB models is a structural change of the whole pension system, it is important to identify political and economic factors that influenced this major policy change. I summarise the existing literature by first looking at determinants of pension reform in general and then focusing on MPP and pension privatisation.

When it comes to pension reform in general, numerous factors have been analysed. On the topic of privatisation, various academics have highlighted that partisan politics and the power of left-wing parties are important factors (Doctrinal, 2023). In Europe, trade unions play a key role in determining the success of proposed reforms (Schneider, 2009), but factors such as constraints on the executive, political competition, fiscal federalism, and regional autonomy are also significant (Verbič and Spruk, 2019). On the economic side, business cycles have proven to have a significant impact on reforming pension systems (Beetsma et al., 2017).

The adoption of MPPs and pension privatisation, in general, has also been an area of interest for economists and political scientists. The main demographic factor examined, which also served as the World Bank’s justification for pension pillarisation, is the sustainability challenge posed by ageing populations. In such countries, privatisation can reduce the financial burden on the state budget (James, 1998). The most widely discussed determinant is the influence of the World Bank as the leader of a transnational campaign advocating for MPP reforms (Orenstein, 2008b). This is especially relevant in the CEE region, where two region-specific factors are present: policy diffusion among neighbouring countries (Adascalitei and Domonkos, 2018) and the desire to ‘out-liberalise’ the EU during the accession process to show leadership potential in policymaking (Orenstein, 2008a, p. 911). Beyond politics, macroeconomic factors are also seen as an important determinant for MPPs (Müller, 2014).

Most of the research done on MPP in Europe discusses potential determinants, issues, and outlooks by comparing certain factors among countries that adopted an MPP (Dobronogov and Murthi, 2005; Wagner, 2005). The qualitative literature primarily focuses on specific CEE countries, often through case studies or comparisons between a few select states (Adascalitei, 2017; Adascalitei and Domonkos, 2015). Common factors influencing the adoption of MPPs identified by these analyses include the difficulty of transitioning to a capitalist economy in the former socialist states and the influence of the World Bank model (Adascalitei and Domonkos, 2015; Iwasaki and Sato, 2008).

In terms of quantitative research, the existing literature is scarce. One of the most comprehensive studies identified focuses predominantly on economic determinants, using accelerated failure time models across a sample of ninety-six countries (Fong and Leibrecht, 2020). Their main findings are that extreme inflation and globalisation shorten the time for MPP adoption, while World Bank loans and legislative fractionalisation do not have a significant influence.

The literature review clearly shows an imbalance in research employed for MPP adoption: while there are plenty of qualitative studies, quantitative approaches have not been widely used. Moreover, there is no quantitative study focusing specifically on political factors. As such, there is a clear research gap when it comes to the quantitative study of political determinants of MPP adoptions.

3. Theoretic Approach – Political Ideology And Transnational Actors

To analyse the role of political factors that influence the adoption of a mandatory private pillar, I first turn to theories of political economy to better understand the relationship between politics and welfare policy choices. I approach the topic on the domestic and international level. For the former, I look at the role of party ideology through the lens of Partisan Theory (Hibbs, 1977) and its application in pension reform, as developed by Yesola Kweon and Kohei Suzuki (2022). For the latter, I refer to the role of transnational actors in shaping national policy, as theorised by Mitchell A. Orenstein (2008b).

3.1. Political Ideology And Pension Reform

Partisan Theory examines the role of political parties and their ideology in shaping public policy. While first used to explain differences in macroeconomic policy (Hibbs, 1977), the theory was expanded for usage in other economic policy areas as well (Hibbs, 1992). In simpler terms, the policy preferences of a party are shaped by the socioeconomic position of its core supporters. This perspective is similar to the Selectorate Theory, which holds that political leaders, driven by the desire for re-election, will implement public policies to maintain the support of the winning coalition—the share of voters essential for securing re-election (De Mesquita et al., 2005).

Transposed to the party sphere, the socioeconomic group supporting the party influences its policy perspective, and the party will advocate for or implement policies to maintain the support of the socioeconomic group. Traditionally, left-wing parties enjoyed the support of the working class, while right-wing parties advocated for positions to favour the middle and upper classes. Taking welfare policies as an example, left-wing parties tend to support welfare expansion and redistributive policies, while the right generally would advocate for the maintenance or reduction of social spending (Esping-Andersen, 1990; Korpi, 1989).

There are two aspects to be considered when applying the Partisan Theory to modern pension reforms. First, newer versions of Partisan Theory go beyond the classic left-right economic divide, as modern societies tend to become more fragmented when it comes to political demand (Bonoli, 2007). With class no longer representing the main cleavage to organise political demands around, parties have to reshape their ideological demands to fit the transformation of the society and implicitly of the electorate (Häusermann et al., 2013). While this is an important distinction, economic policy is still an area where distinctions among socioeconomic groups can explain the ideological positions of political parties (Kweon and Suzuki, 2022).

Additionally, pensions represent a special policy area even within welfare policy as a whole, as all citizens have an incentive to support generous pension programmes. All individuals, regardless of socioeconomic position, will eventually receive old-age payments. As such, there is an overall interest in maintaining adequate pension policies, even among richer and younger individuals (Jensen, 2012; Laenen, 2020). This, in turn, incentivises all parties, irrespective of ideology, to support such policies (Breyer and Craig, 1997). However, parties across the political spectrum can still be differentiated by the policy tool for which they are advocating (Zehavi, 2012). While the goal of offering generous pensions remains constant across all parties, the way to achieve this differs based on ideology.

In this regard, the policy tools can be understood as the choice between market- or state-oriented solutions (Kweon and Suzuki, 2022). This choice is particularly influenced by party ideology, which determines one’s perspective on the role of the state in the economy (Strom, 1990), especially with respect to the topic of privatisation (Engler and Zohlnhöfer, 2019). Right-wing parties tend to envision a reduced role for the state in society and advocate for a mix of state regulation and market-based solutions. In the area of welfare, this means advocating for the integration of market-based social services (Kweon and Suzuki, 2022). The MPP represents a moderate form of privatisation, as it does not completely replace the DB PAYG pillar but rather complements it. This offers right-wing parties the opportunity to achieve two goals simultaneously: it shows support for generous pensions by retaining the Bismarckian first pillar with the aim of optimising the pension system to address challenges posed by an ageing population while advocating for more market-based solutions. Compared to other forms of pension privatisation, the MPP is the more balanced policy option: it does not create a new system like full privatisation would but also goes further than simply offering tax benefits, as seen in voluntary private pillars.

On the other hand, left-wing parties have different ideological perspectives. The envisioned role of the state is much greater, especially in the area of social protection and welfare (Kweon and Suzuki, 2022). One guiding principle of left-wing parties is equality, and the economic means to achieve it are redistributive policies (Zehavi, 2012). Applied to the pension systems, left-wing parties will, ideologically speaking, prefer a PAYG DB benefit pillar over private pillars. The reason is that the Bismarckian pillar is redistributive to some degree by design, while the MPP is a defined-contribution (DC) system, where pensions are solely determined by the amount stored by the individual in their pension account and the investment returns achieved by the fund. This would mean that a left-wing party is expected to choose a Bismarckian pillar over pension privatisation, but pillarisation means having the two systems work concomitantly and complementarily. In this case, the choice can be determined by two factors: the MPP is usually created by diverting a portion of the funds to be paid to the first pillar, thus weakening it. Additionally, even if it produces less inequality than full pension privatisation, a multi-pillar system can be expected to produce more inequality among the elderly in comparison to the classic Bismarckian pension. Private pensions are, in general, associated with higher income inequality among the elderly (Been et al., 2017). As such, left-wing parties should be against, or at least sceptical about, the introduction of an MPP.

Drawing on Hibbs’ (1977) partisan theory and its application to pension reform by Kweon and Suzuki (2022), I hypothesise that:

H1: Having a right-wing party in government increases the probability of a country adopting a mandatory private pension pillar.

3.2. Transnational Actors And Pension Reform

While the political ideology of the government can theoretically explain the role domestic politics plays in pension reform, the picture is incomplete without considering influences at the international level. As shown, the adoption of MPPs did not occur in isolated cases but was instead part of a broader reform trend that spread across the world. Moreover, international organisations, especially the World Bank, played a significant role in the debate surrounding MPPs, advocating for the pillarisation of pensions. To provide more theoretical context, I refer to Orenstein’s model of transnational actor influence (Orenstein, 2008b), which highlights the role of transnational coalitions across three stages of the policy process.

Adding to the veto players model (Tsebelis, 2002), Orenstein posits that since transnational actors are not embedded in the political systems of individual nations, they take on the role of ‘proposers’ (Orenstein, 2008b). As such, the first steps when advocating for a certain policy are domestic agenda-setting and coalition-building. The transnational actor itself can overcome its lack of domestic power by mobilising resources to find and lobby domestic allies who have veto power. Thus, the actual power of transnational actors is the number of resources they can mobilise. These can take on different forms and functions, from workshops and conferences meant to promote the desired policy to international loans to ensure the monetary requirements of the policy. Additionally, transnational actors often work not alone but in coalitions (Tarrow, 2005). The resources of actors or coalitions are employed over time, as policy advocacy is not a singular event but a process. As such, the model identifies three different stages of policy adoption: policy development, policy transfer, and implementation.

The first stage consists of the emergence of a policy model and its first implementation (Orenstein, 2008b). Transnational players act as an epistemic community (Haas, 1992) that collates similar views regarding political and economic problems to find a solution. Ideas themselves can be the common denominator that shapes international coalitions (Béland and Cox, 2016). The country where the policy is first adopted plays a pivotal role, as it is a policy laboratory where the proposed ideas first take shape and are implemented, highlighting possible difficulties or other aspects that were not visible before (Bockman and Eyal, 2002). When it comes to pension privatisation, the policy laboratory was Chile, where pension privatisation first took place as part of the broader economic liberalisation of Chile, advocated for by the ‘Chicago boys’ (Valdés, 1995). Thus, the reform was the implementation of economic ideas formed within the neoliberal epistemic community advocating for pro-market policies and monetarism (Orenstein, 2008b). The MPP model was elevated to a new level of international influence when the World Bank publicly advocated for pension pillarisation, as explained in the previous section. When it comes to the World Bank’s advocacy, there are two important aspects to consider: it represented the successful recruitment of the World Bank in the transnational coalition while conferring legitimacy and reputability to MPP policies.

Following initial adoption, the second stage of ‘policy transfer’ consists of the advocacy for policy adoption in other countries (Orenstein, 2008b). The transnational coalition supports this policy diffusion through soft and hard power (Brooks, 2005). While the former form of power encompasses activities such as conferences, seminars, and study trips, the latter includes more direct forms of coercion such as loans and technical assistance (Orenstein, 2008b). In the case of MPPs, the policy diffusion activities differ from country to country but tend to follow a pattern: first, the World Bank would find potential domestic partners through international conferences or seminars held at prestigious universities (Orenstein, 2008b). This is followed by building partnerships with the domestic actors and providing them with resources to allow for successful policy adoption. In this case, resources also include technical assistance, which plays a major role after the political willingness is secured (Hinz and Holzmann, 2005). The most important leverage of financial organisations involved in the coalition is loan conditionality. The World Bank did provide loans on the condition that recipient countries implement a form of privatisation, including MPPs (Orenstein, 2008b). The loan itself, seen as proof of economic performance and a possible positive signal to investors (Orenstein, 2008b), can be an incentive for domestic actors to implement the policy (Müller, 2003).

The last stage is the implementation of the desired policy. While the transnational coalition reaches its first goal when the policy is adopted, it has an incentive to ensure a successful implementation in the country, as this can influence the chances of further policy diffusion (Orenstein, 2008b). Moreover, the transnational coalition can convince national players to adopt the policy by promising involvement and assistance throughout the implementation process, thus minimising the national actor’s apprehension of implementation difficulties or unsuccessful implementation. In the case of MPPs, this function is shared between the World Bank and USAID (Orenstein, 2008b). While the former provides the loans, continuous implementation assistance is ensured through the latter. Formal policy adoption is thus only the first step and is significantly less resource-consuming than the actual implementation.

While I applied the theoretic framework outlined by Orenstein directly to the World Bank, this does not mean that other pension reforms cannot be explained by applying the model to other transnational actors. One noteworthy example is the European Union, which successfully influenced Hungary and Greece to stop or reverse pension privatisation through loan conditionality during the sovereign debt crisis (Stepan and Anderson, 2014). While these events come after the popularity of pillarisation had subsided, they highlight an important insight: The EU, or transnational actors in general, can effect change mainly through ‘hard power’, understood here as providing conditional monetary means to encourage the adoption of certain policies. Accordingly, as the World Bank has major monetary means it can mobilise, I expect its influence to be more visible than that of other transnational coalitions when it comes to influencing the adoption of MPPs.

Taking into account all advocacy forms transnational coalitions can take across all three stages, international actors can play a major role in national pension reform. The World Bank, as the most highly regarded institution among the coalition members and with a large amount of resources, seems to have a crucial role in the diffusion and implementation stages. The ability to provide technical assistance and, arguably more importantly, to provide loans to countries if they adopt a form of pension pillarisation makes it the most visible actor on the international stage in pension reform. As such, I define my second hypothesis as:

H2: Greater World Bank involvement increases the probability of a country adopting a mandatory private pension pillar.

4. Quantitative Analysis – MPP Adoption In The Eu

To empirically analyse the role of right-wing parties in government and the influence of the World Bank, I first approach the topic from a quantitative perspective. I focus my analysis on MPP adoption in the current twenty-seven countries of the European Union.

4.1. Data And Method

Employing a comparative perspective, the data used for this analysis is structured as country-year. As mentioned, I take the current EU countries as my analysis sample. It is important to note that the observation time frame differs from country to country based on the state of the political system in each country at various points in time. While the general starting year for observation is 1980 (the year Chila first privatised their pensions), this may not be a suitable starting point for all countries within the dataset. For instance, the CEE countries were yet to transition from communism to liberal democracy. As such, for these countries, I take the year after the regime change or independence as the first observation year. This means 1990 for most former Eastern Bloc countries, 1992 for former Yugoslav countries, 1992 for Baltic countries, and 1993 for Czechia and Slovakia. The observed time frame ends with the year of adoption or 2020 if the country did not adopt an MPP.

Following the established methodological approaches for policy adoption analysis, I use binary time-series cross-section (BTSCS) and survival models. The former consists of a pooled logit with robust estimations of variance following research published on the topic of civil war onset (Bussmann and Schneider, 2007). Additionally, in concordance with the specific considerations for BTSCS models, I add year dummies (Beck et al., 1998). The latter model employs extended Cox regressions to accommodate time-dependent variables, following the recommendations of the literature (Jones and Branton, 2005; Singer and Willett, 2003).

DEPENDENT VARIABLE

Adoption of an MPP pillar – As usual in BTSCS analysis, the dependent variable is a dummy that holds for the year a country has adopted the policy and 0 otherwise. I use the data provided by Fong and Leibrecht (2020) in their analysis of MPP adoption worldwide.

INDEPENDENT VARIABLES

Right-Wing Government Seat Share – This variable follows the first hypothesis I developed and measures the relative power of right-wing parties through the percentage of ministerial seats held by parties considered right-wing. I use the data from the Comparative Political Dataset (Armigeon et al., 2024).

World Bank Pension Project – As the influence of the World Bank in the area of pensions is hard to isolate when considering total World Bank loans to a country, I use a dummy to verify whether the World Bank has at least one active project related to pensions in a given year. I coded this variable by using the available data on the World Bank projects provided on the official website. I consider a project to be related to pensions if it has the label ‘Social Insurance and Pensions’.

CONTROL VARIABLES

New Government Ideology – As the observation time frame is large and the countries observed have democratic regimes, there is the possibility of government change due to periodical elections. This dummy holds if there is a change in government ideology. I expect a change in government ideology to harm the probability of MPP adoption, as new governments need time to formulate and adopt new policies. The data is taken from the Comparative Political Dataset (Armigeon et al., 2024).

Government Support – Pension reforms require a legislative process, meaning that the government needs a stable majority in Parliament to adopt an MPP. As MPPs can be controversial, a larger majority might be needed to ensure adoption. This variable measures the parliamentary seat share of all parties in government. The data is taken from the Comparative Political Dataset (Armigeon et al., 2024).

Stocks traded – One of the factors influencing pension privatisation is financialisation (Braun, 2022; Jackson, 2013; Natali, 2018). As privatised pension funds can be invested in the financial market, governments might want to promote financial growth through pension privatisation. This variable measures the total value of stocks traded as a percentage of GDP. The data is taken from the World Bank (World Bank, 2023d).

Market capitalisation – This variable measures the market capitalisation of domestic companies as a percentage of GDP. The data is taken from the World Bank (World Bank, 2023b).

Old-age dependency ratio – The main justification offered by the World Bank for pension pillarisation is population ageing. Much of the debate surrounding new pension systems has been centred on the unsustainability of the Bismarckian system in an ageing society (Orenstein, 2013). This variable measures the ratio of people older than sixty-four to the working-age population. I expect it to be positively correlated to the probability of MPP adoption. The data is taken from the World Bank (World Bank, 2023a).

Population growth – This variable also looks at ageing by measuring the population growth rate for each year. I expect negative growth to increase the chances of pension privatisation. The data is taken from the World Bank (World Bank, 2023c).

Annual deficit – The remaining variables control the economic state of the country. Research has shown that deficit can be a determinant of pension privatisation (Poortvliet and Laine, 1995). The data measures the annual deficit as a percentage of GDP and is taken from the Comparative Political Dataset (Armigeon et al., 2024).

Openness of the economy – Especially in the context of Europe, research has shown that some countries adopted MPPs to reflect an adherence to the liberal principles of the EU (Orenstein, 2008a). MPPs might be more easily adopted in countries where neoliberal principles, like economic openness, have a stronger influence. This variable measures the sum of exports and imports as a percentage of GDP and is taken from the Comparative Political Dataset (Armigeon et al., 2024).

Inflation – Research has shown that extreme inflation can lead to pension privatisation (Fong and Leibrecht, 2020). The data is taken from the Comparative Political Dataset (Armigeon et al., 2024).

The last three variables (Social Security transfers, Social expenditure, and Old-age expenditure) also look at the burden of pensions on the state budget and are taken from the Comparative Political Dataset (Armigeon et al., 2024).

4.2. Analysis

Table I presents descriptive statistics for the observation timeframes, main independent variables, and dependent variables.

	Country	First Obs.	Last Obs.	Ideology	WB Project  Start	WB Project  End	MPP  Adoption
1	Austria	1980	2020	11.31	–	–	–
2	Belgium	1980	2020	34.32	–	–	–
3	Bulgaria	1990	2002	47.07	1996	2002	2002
4	Croatia	2000	2002	34.93	2002	2002	2002
5	Cyprus	1980	2020	52.94	–	–	–
6	Czechia	1993	2020	43.82	–	–	–
7	Denmark	1980	2020	62.19	–	–	–
8	Estonia	1992	2002	40.79	–	–	2002
9	Finland	1980	2020	36.49	–	–	–
10	France	1980	2020	34.01	–	–	–
11	Germany	1980	2020	12.62	–	–	–
12	Greece	1980	2020	35.98	–	–	–
13	Hungary	1990	1998	42.11	1998	1998	1998
14	Ireland	1980	2020	55.14	–	–	–
15	Italy	1980	2020	22.16	–	–	–
16	Latvia	1993	2001	73.75	1997	2001	2001
17	Lithuania	1992	2004	22.11	2000	2004	2004
18	Luxembourg	1980	2020	18.67	–	–	–
19	Malta	1980	2020	0	–	–	–
20	Netherlands	1980	2020	27.31	–	–	–
21	Poland	1991	1999	30.94	–	–	1999
22	Portugal	1980	2020	45.68	–	–	–
23	Romania	1990	2008	26.41	1995	2008	2008
24	Slovakia	1993	2005	55.77	2002	2005	2005
25	Slovenia	1993	2020	46.48	–	–	–
26	Spain	1980	2020	0	–	–	–
27	Sweden	1980	2007	14.77	–	–	2007

Table I – Descriptive Statistics

The Ideology columns indicate the mean government seat share of right-wing parties for the whole observation period. The data itself is diverse, ranging from zero per cent for Malta and Spain to 73-75 % for Latvia. In terms of World Bank presence, seven of the observed countries had pension-related World Bank projects: Bulgaria, Croatia, Hungary, Latvia, Lithuania, Romania, and Slovakia (World Bank, 1996, 1997a, 1997b, 2000a, 2000d, 2001a, 2001b). All of them are CEE countries, and the projects started in the mid to late 1990s or early 2000s. The end date represents the year the project finished or the year the MPP was adopted. In all seven cases, the MPP was adopted before the World Bank projects were finished. In addition to the seven countries, three other states adopted an MPP in the observation period. Of the total ten, nine are from CEE, and Sweden is the only Western European country that has opted for such a policy. All the MPPs were adopted at the end of the 1990s and the beginning of the 2000s, with Poland and Hungary as the first ones and Romania as the last adopting country.

Table II shows the regression results for the pooled logit with robust estimation of variance models. The first model only includes the two main independent variables, and the next models add control variables that are related to each other: government variables in model 2, financialisation in model 3, sociodemographic in model 4, economic variables in model 5 and expenditure-related variables in the last model. Model 2 is the only one to include variables related to financialisation, as the data had numerous missing observations, resulting in an N of 550 instead of 791.

	Model1	Model2	Model3	Model4	Model5	Model6
Intercept	-23.31**	-24.02***	-23.17	-24.11**	-25.22***	-19.73
	(8.92)	(3.44)	(56.63)	(8.48)	(5.08)	(13.69)
Right-Wing Seats Share	0.02*	0.02*	0.03	0.02	0.02*	0.03**
	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)	(0.01)
World Bank Pension Project	3.26***	3.34***	2.89	2.58	2.60	0.27
	(0.77)	(0.90)	(2.50)	(1.66)	(1.65)	(1.81)
New Government Ideology		-0.69	-20.03***	-0.66	-0.59	-0.03
		(1.03)	(1.50)	(1.07)	(1.03)	(1.45)
Government Support		0.01	-0.03	0.02	0.02	0.05
		(0.03)	(0.07)	(0.03)	(0.03)	(0.03)
Stocks traded % GDP			-0.07
			(0.04)
Market Capitalization			-0.01
			(0.04)
Old-age Dependency Ratio				0.01	0.04	-0.12
				(0.15)	(0.20)	(0.21)
Population Growth				-0.58	-0.67	-2.57**
				(0.66)	(0.63)	(0.88)
Annual Deficit					-0.02	0.25*
					(0.16)	(0.11)
Openness of the Economy					0.01	0.00
					(0.01)	(0.01)
Inflation						-0.13
						(0.11)
Social Security Transfers						0.35
						(1.14)
Social Expenditure						-0.20
						(0.69)
Old-age Expenditure						-0.44
						(0.79)
AIC	139.60	142.89	110.91	145.79	147.50	137.70
BIC	340.93	353.19	313.47	365.44	366.70	371.61
Log Likelihood	-26.80	-26.44	-8.45	-25.90	-25.75	-16.85
Deviance	53.60	52.89	16.91	51.79	51.50	33.70
Num. obs.	798	791	550	791	711	664

Table II – Logit Models

		Model1		Model2		Model3		Model4		Model5		Model6
Right-Wing Seats Share		0.01*		0.01*		0.03**		0.01*		0.02**		0.02*
		(0.01)		(0.01)		(0.01)		(0.01)		(0.01)		(0.01)
World Bank Pension Project		2.94***		2.93***		1.83		2.44		2.34*		1.85
		(0.84)		(0.84)		(1.92)		(1.26)		(1.12)		(1.19)
New Government Ideology				-0.05		-17.86***		0.03		-0.06		-0.40
				(0.71)		(1.03)		(0.77)		(0.77)		(1.29)
Government Support				-0.01		-0.02		-0.02		-0.02		-0.02
				(0.02)		(0.03)		(0.02)		(0.02)		(0.03)
Stocks traded % GDP						-0.15
						(0.09)
Market Capitalization						0.02
						(0.02)
Old-Age Dependency Ratio								0.02		0.07
								(0.12)		(0.17)
Population Growth								-0.38		-0.45		-0.91
								(0.53)		(0.44)		(0.57)
Annual Deficit										-0.02		0.19
										(0.14)		(0.13)
Openness of the Economy										0.01		0.00
										(0.01)		(0.01)
Inflation												-0.12*
												(0.06)
Social Security Transfers												-0.16
												(0.55)
Social Expenditure												0.02
												(0.32)
Old-Age Expenditure												-0.04
												(0.22)
AIC		59.51		63.18		14.97		61.81		63.91		40.36
R2		0.02		0.02		0.03		0.02		0.03		0.03
Max. R2		0.07		0.07		0.05		0.07		0.08		0.06
Num. events		10		10		5		10		10		7
Num. obs.		798		791		550		791		711		664
Missings		11		18		259		18		98		145
PH test		0.29		0.51		0.58		0.76		0.69		0.68

Table III – Extended Cox Models

The governmental seat share of right-wing parties has a significant effect on the adoption of an MPP in four out of the six models. It only loses significance when controlling for financialisation and sociodemographic factors, but the coefficient is consistent and positive across all models.

On the other hand, the influence of the World Bank, measured through its pension projects, is only significant in the first two models. Nonetheless, the coefficient is consistent, as it is positive across all models. Only in the last model, which adds the expenditure-related variables, does the coefficient drop in comparison to all previous models.

The government-related control variables are added beginning with the second model, with ambiguous results. While the new ideological composition of the government is consistent with the expectations, the government support variable is inconsistent and changed the sign to negative, but only in the third model. However, the new government ideology is only significant in the third model, and the government support does not reach significance in any model. This means that there is only limited evidence to show that a new government with a new ideology decreases the chances of MPP adoption. There is also no evidence to support the assumption that governments with larger majorities (Government Support) are more likely to adopt MPPs.

For the financialisation variables, both are negative and in line with theoretical expectations, but none are significant. The results for the sociodemographic variables are also ambiguous. The coefficient for the old-age dependency ratio goes in line with expectations but is not significant. On the other hand, the population growth is negative but significant only in the last model. It shows that countries with higher population growth tend not to adopt an MPP, or inversely, a country with lower or negative population growth tends to opt for pension pillarisation, in line with the theoretical considerations. Other than that, there is no empirical support for financialisation influencing MPP adoption. The economic and expenditure-related variables do not reach statistical significance, and in the case of the annual deficit, the results are also not consistent, although the variable is significant in the last model. Because of this, a conclusion cannot be reached.

Table III introduces the results of the survival models. The models follow the pattern of the logit regressions, adding one block of control variables with each new model. As the Cox regression measures the hazard ratio, a positive coefficient means a higher risk of MPP adoption, and a negative one means a lower risk.

For the government share of right-wing parties, the results are positive, consistent, and significant across all models. The World Bank pension projects variable is consistent but only significant in the first two models and the fourth model. This shows that both a higher share of right-wing ministers and the existence of a pension project increase the risk of adopting an MPP, which is in line with the theoretical expectations.

The variables for government, financialisation, economy, and expenditure are insignificant across all models, with one exception. For new government ideology, the variable is significant in the third model, but the results are not consistent, as the coefficient fluctuates from negative to positive across models. Only inflation reaches significance with a negative coefficient, contradicting the findings of previous research, which showed it having no influence on MPP adoption (Fong and Leibrecht, 2020).

The sociodemographic variables are ambiguous, similar to the logit models. The old-age dependency ratio is insignificant but consistent. However, the population growth is negative but significant across all models. One interesting insight is that none of the sociodemographic factors are significant, although they are usually thought about as the main determinant for MPP adoption. It seems that economic considerations, such as the unsustainability of Bismarckian pensions in ageing societies, are not factors influencing pension pillarisation in the models.

Therefore, the results from all models paint a clear picture: political considerations have a significant influence on the adoption of MPPs. There are statistical clues that both right-wing governments and the influence of the World Bank led to the adoption of an MPP. While both variables were insignificant in some of the models, they both were consistent in their direction – the variables exhibit some fluctuation but do not change signs across models, showing consistency even when control variables were added. Only the World Bank variable lost significance and presented increased standard errors when confronted with the series of control variables I employed. Right-wing seat share was significant in all Cox models and in the majority of Logit models, and the World Bank project was significant in almost half of the models. As such, while the results for the World Bank are too ambiguous to draw definitive inferences, the situation regarding right-wing seat share is clear. As such, in the case of I can reject the null hypothesis, but in the case of I fail to do so. Therefore, the findings of the quantitative analysis support , while there is inconclusive evidence to support .

5. Qualitative Analysis – The Case Of Romania And Beyond

To investigate the role of right-wing governments and the World Bank’s influence on the adoption of MPPs further, I approach the topic from a qualitative perspective, using the case of Romania. There are several reasons for selecting Romania: first, it is one of the seven EU countries that had a World Bank pension project before opting for pension pillarisation. Additionally, it changed governments during the adoption process, from a left-wing social-democratic one to a centre-right one in 2004, thus providing enough information and variation to look at the ideological influences on pillarisation of both left- and right-wing parties in government.

It is important to note that this second empirical analysis is meant to illustrate and support the findings of the previous quantitative study because it allows for a better understanding of the mechanisms and details that can be lost in quantification. Accordingly, the case study does not aim to reject the null hypothesis and draw inferences about the population of cases, as I already did in the previous section.

To augment the insights drawn from the case of Romania, I also look at the main elements of other World Bank pension projects in Bulgaria, Hungary, Lithuania, Slovakia, and Croatia. I focus on the role of the Bank, as its project information is readily available in English and can be verified, unlike the political positions of the respective governments. The role of this second analytical part is to compare the results in the case of Romania, highlighting differences and other possible mechanisms through which the World Bank projects influence MPP adoptions that are not visible in the Romanian case study. The focus here lies on the situation before and after the pension project emerged and finished, thus exploiting the within-case variation to isolate the role of the World Bank.

Ideology related documents
	Title	Type	Year	Source
1.	Program de guvernare pe perioada 2001-2004	Government program	2001	Monitorul Oficial
2.	Expunere de motive – PL-x nr. 424/2004 (Lege 411/2004)	Exposition of motives	2004	Camera Deputatilor
3.	Program de guvernare pe perioada 2005-2008	Government program	2004	Monitorul Oficial
World Bank related documents
	Title	Type	Year	WB Report No.
1.	Romania – Social Sector Development Loan Project (English)	Project Information Document	2001	PID9325
2.	Romania – Social Sector Development Project (English)	Project Appraisal Document	2001	22130
3.	Statement by Balmiki Prasad Singh at the meeting of June 19, 2001 (English)	Executive Director’s Statement	2001	87074
4.	Romania – Social Sector Development Project (English)	Implementation Completion and Results Report	2009	ICR1042
5.	Romania – Social Sector Development Project (English)	Implementation Completion Report Review	2009	ICRR13232

Table IV – Documents for the Case Study of Romania

	Title	Type	Year	WB Report No.
	Bulgaria – Social Insurance Administration Project (SIAP)
1.	Bulgaria – Social Insurance Administration Project (English)	Staff Appraisal Report	1996	15531
2.	Bulgaria – Social Insurance Administration Project (English)	Implementation Completion and Results Report	2002	251130
3.	Bulgaria – Social Insurance Administration Project (siap) (English)	Implementation Completion Report Review	2003	ICRR11429
	Latvia – Welfare Reform Project
4.	Latvia – Welfare Reform Project (English)	Project Appraisal Document	1997	16487
5.	Latvia – Welfare Reform Project (English)	Implementation Completion and Results Report	2004	29347
	Hungary – Public Sector Adjustment Loan (PSAL)
6.	Hungary – Public Sector Adjustment Loan Project (English)	Project Information Document	1997	PID5813
7.	Hungary – Public Sector Adjustment Loan Project (English)	Implementation Completion and Results Report	2000	20523
	Lithuania – Structural Adjustment Loan 2 (SAL 2)
8.	Lithuania – Second Structural Adjustment Loan Project (SAL II) (English)	Project Information Document	2000	PID8801
9.	Lithuania – Second Structural Adjustment Loan Project (English)	President’s Report	2000	P7387
10.	Lithuania – Second Structural Adjustment Loan Project (English)	Implementation Completion and Results Report	2003	25667
	Slovakia – Social Benefits Reform Administration Project
11.	Slovak Republic – Social Benefits Reform Administration Project (English)	Project Information Document	2001	PID8058
12.	Slovak Republic – Social Benefits Reform Administration Project (English)	Project Appraisal Document	2002	22894
13.	Slovak Republic – Social Benefits Reform Administration Project (English)	Implementation Completion and Results Report	2008	ICR810
	Croatia – Pension System Investment Project
14.	Croatia – Pension System Investment Project (English)	Project Information Document	2000	PID9384
15.	Croatia – Pension System Investment Project (English)	Project Appraisal Document	2002	23714
16.	Croatia – Pension System Investment Project (English)	Implementation Completion and Results Report	2009	ICR1037

Table V – Documents for the Seven World Bank Projects

5.1. Data And Method

In the first part of my qualitative analysis, I isolate and analyse the effect of the government ideology and World Bank influence by looking at two sets of documents. For the role of the government, I look at the government programmes of the Năstase (2000-2004) and Tăriceanu (2004-2008) cabinets, specifically at the sections regarding pensions and social insurance. Additionally, I also analyse the exposition of motives the government provided to parliament when it proposed the law project that would later become the main framework for the MPP – Lege 411/2004. For the role of the World Bank, I look at all publicly available and relevant documents from the Romania Social Sector Development Project and the corresponding loan of the World Bank, using the publicly released material on the World Bank website. Table IV summarises all the documents used for the qualitative analysis.

For the second part of the qualitative analysis, I examine only two sets of documents for each case: the project information and appraisal documents, as well as the implementation completion and results report. For the former, I focus specifically on information regarding the situation before the conception of the project, while for the latter, I look for clues regarding the effective role of the World Bank project and the situation after its finalisation. I can thus compare the status of MPP legislation and implementation before and after the involvement of the World Bank, isolating its role. Table V provides a summary of all the documents used in this second part of the qualitative analysis.

I analyse the documents using the content analysis method. As such, I do not discuss the details and technicalities of the document at all, but only the relevant sections related to my research question, and especially to the main independent variables. For the government ideology, the goal is to determine the position of the governing parties on MPP. For the World Bank, I examine the project documents to explain the role of the bank in MPP adoption.

5.2. A Case Study Of Romania

The process through which Romania adopted its MPP is long and takes place from the beginning of the 2000s until 2008. While the government attempted to privatise pensions quickly through emergency ordinances, these efforts ultimately failed. The MPP was adopted through a gradual process, beginning with the Lege 411/2004, which established the necessary framework and was initially planned to be fully functional by 2006. However, due to the need for further legislative decisions and the delay in the adoption process, the system was only fully adopted and functional in 2008.

The left-wing Năstase cabinet took power in 2000 with a government programme that set out goals to introduce a vast reform of the whole pension system (Monitorul Oficial, 2000). While the document does not mention the World Bank model, the core concept was a social insurance system based on pillars. The public pillar was considered the foundation, and the cabinet proposed to optimise it (Monitorul Oficial, 2000). In addition, a mandatory private second pillar would be introduced once there were certainties regarding the coverage of the budgetary deficit it would create. The programme sharply criticises the previous attempts to privatise pensions at this point. Additionally, the programme also mentions a further pillar that resembles voluntary occupational pensions (Monitorul Oficial, 2000).

Close to the end of its term, in 2004, the left-wing government introduced the MPP law project in parliament. In the exposition of motives, it mentions the difficulties in raising pensions due to the ageing population and the unsustainability of the previous system due to the same reasons as motivations for the introduction of this law project (Guvernul României, 2004). The document repeats and sets out the three-pillar structure: a first public pillar, a second mandatory private pillar, and a third voluntary occupational pillar. While the main reasons mentioned throughout the document were ageing and sustainability, additional motives were the need to provide capital infusions in the internal financial market and policy-learning from EU and candidate countries (Guvernul României, 2004). The project also included safety measures, such as a reserve fund.

Shortly after the law was passed, a new centre-right government took office. In its government programme, it only briefly mentions the MPPs and the goal to improve its legislative framework and bring it to a functional form (Monitorul Oficial, 2004). The overall goals regarding pension reform are poverty reduction among the elderly and contribution-based pensions.

There are clear differences between the government perspectives between the two governments. The social-democratic perspective clearly shows that its ultimate goal is raising pension levels and also presents scepticism to the MPP until the deficit it would create is covered. The centre-right cabinet also shares the goal of poverty reduction, but it advocates for contribution-based pensions, which can be interpreted as a right-wing perspective. However, even if these small ideological details are present, both cabinets chose to support an MPP.

Concomitantly to this development in domestic politics, the World Bank was active in the country through the Social Sector Development (SSD) project, which started in 2001 and was meant to help reform Romania’s social sector through four main components: policy development for the Ministry of Labor, pension reform, labour market adjustment, and social assistance strengthening. While the documents publicly disclosed by the bank are rather technical, they show the role of the project in MPP adoption.

The goals of the SSD in pension reform were to help the government introduce a multi-pillar structure (World Bank, 2001a). While it praised the former reforms for public pensions, it still considered the PAYG system unsustainable without an MPP (World Bank, 2001a). The project also outlines the need for strict regulations for an effective MPP and the deficit it would create in the first years after adoption. Thus, the successful adoption and implementation were dependent on strengthening the financial sector and stabilising the economy. The World Bank plays multiple crucial roles: together with the EU and IMF, it endorsed the SSD with MPP as a component. Additionally, it was the largest lender for social protection programmes and had contacts within the Romanian government, and it assisted based on its experience with MPP adoption in other countries (World Bank, 2001a).

In a statement made during a meeting shortly after the project was launched, a World Bank official praised Romania’s efforts to reform its system after the legacy of the communist regime (Singh, 2001). The main points were the need for further privatisation, not just in the pension sector, but regarding multiple state-run enterprises. Regarding pensions specifically, the statement welcomed the reform project but criticised the project for not having concrete strategies to create support for pillarisation within the population to avoid popular backlash (Singh, 2001). One important insight of this document is the Bank’s perspective that the EU should lead the agenda-setting for development policy.

At the end of the project, the Independent Evaluation Group assessed the performance of both the bank and the government. Regarding pension pillarisation, the World Bank assisted in creating the regulatory framework, evaluating it as ‘sound’ (Independent Evaluation Group, 2009). The financial performance of the MPP was assessed as strong in comparison to other countries. However, the overall Bank performance was only moderately satisfactory, as the main performance was utilising its operational experience, but it did not take into account the complexity of the project and overestimated the government’s capacity, leading to delays in adoption (Independent Evaluation Group, 2009). On the implementation side, the Bank was more responsive, supervising and tracking the process and outputs.

Overall, the publicly available documents from the SSD project show the World Bank’s clear involvement in Romania’s adoption of the MPP. While they do not make clear what influence the Bank had in convincing the Romanian government to opt for pension pillarisation, it shows the major role in the actual policy-making and implementation part through consultancy and loan provision.

5.3. Beyond Romania: World Bank Pension Projects In Comparison

Going beyond the case of Romania, the World Bank was active in other pension projects in Europe, more specifically in Central and Eastern Europe. As the descriptive statistics of the quantitative part have shown, there were seven pension projects in the observation period. I will shortly summarise the main elements of each of them, highlighting similarities and differences to Romania and also discussing other relevant mechanisms and dynamics if present.

In Bulgaria, the Social Insurance Administration Project emerged at a time of major pension reform. The ruling government, having a relatively stable parliamentary majority, started vast reforms aimed at restoring the financial viability of the public pension system (World Bank, 1996). In this context, the introduction of an MPP was a secondary important goal. When the World Bank project emerged in 1996, the details of such a proposed MPP were not defined yet, but the overall goal existed. Bulgaria finalised and implemented its MPP in 2002, with the World Bank project as the main enabler in this regard. Reviewing the influence of the project, the Bank assisted the MPP adoption mainly by creating institutional capacity (World Bank, 2002a), and on the monetary side, the Bank worked together with other donors such as USAID and the US Department of Labor (World Bank, 2002a). The performance of the Bank was rated as ‘Highly Satisfactory’, especially in the area of institutional development (Independent Evaluation Group, 2003). Compared to Romania, Bulgaria does not show any significant difference: in both cases, the Bank played a vital role in adopting and implementing an MPP once the political will was present, and its main mechanisms were the provision of monetary means and the creation of institutional capacity.

For Latvia, the situation is similar but presents some particularities. At the moment when the Welfare Reform Project by the World Bank emerged in 1998, the country was already undergoing a vast pension reform, including legislation for an MPP, which at the time was supposed to start functioning in 1998 (World Bank, 1997b). The project description itself only briefly mentions pension pillarisation, but the completion report from 2004 talks extensively about pension pillarisation as one of the main outputs of the project and mentions the Bank’s role in developing a vision based on multi-pillar approaches (World Bank, 2004). Similar to Romania and Bulgaria, the World Bank appears to have played a major role in enabling the implementation of an MPP through its knowledge and lending, but the political will for reform was already present when the project started.

In Hungary, the World Bank started its Public Sector Adjustment Programme in 1997. Similarly to the previous cases, the project aimed to help the country reform its public pension system (the first pillar) while also introducing an MPP (World Bank, 1997a). At the start of the project, Hungary implemented a fiscal adjustment and was in the process of transitioning from a centralised to a free market economy, challenged by budget deficits (World Bank, 1997a). The World Bank project was aimed at helping the country succeed in its public sector reform, which was necessary to achieve Hungary’s long-term goal of joining the EU (World Bank, 1997a). The completion report from 2000 shows how the country implemented an MPP shortly after the project started in 1998. The Bank was extensively involved in creating the Hungarian MPP through its expertise, as Bank employees and Hungarian officials met daily for more than one year in the Pension Working Group, and the Bank involved international experts and kept contact with Hungarian stakeholders (World Bank, 2000b). The implementation reports showcase even more clearly than in the previous cases how the World Bank was the main actor in the creation of the MPP, given its knowledge and capacity to coordinate policymaking.

A different story emerges when looking at the case of Lithuania. The Second Structural Adjustment Loan Project (SAL II) by the World Bank started in 2000 and had a much larger scope than pension reform, and the project presentation does not even mention the notion of an MPP (World Bank, 2000d). However, the Bank’s president’s report from the same year discusses the relationship between the country and MPPs, mentioning Lithuania’s reluctance to entertain the idea of pension pillarisation (World Bank, 2000c). Nonetheless, the report mentions an MPP as a necessary part of fundamental pension reform and outlines the clear goal of having a policy package that could set the fundaments for MPP adoption submitted to the Lithuanian legislature by June 2001 (World Bank, 2000c). Three years later, the completion report summarises the results: while the government sent an MPP law to the legislative in October 2001, the Seimas (Lithuania’s parliament) did not pass it (World Bank, 2003). Nonetheless, the Bank still assisted the government in creating adjacent legislation that would allow for easy implementation of an MPP once there is political will (World Bank, 2003). This case presents a very different reality from others: even when there is not political will for MPP adoption, the World Bank still insists on it being a mandatory part of pension reform. This shows that the Bank is not just a technical assistant helping countries implement their desired reforms but also a political actor with its interests and agenda. While the project reports ended in 2003 with an apparent failure of the Bank’s goals, Lithuania did indeed adopt an MPP in 2004.

In 2002, the World Bank started its Social Benefits Reform Administration Project, helping Slovakia reform its pension system, which was experiencing a deficit (World Bank, 2001b). Due to macroeconomic reasons and especially demographic change, the long-term sustainability of its pension system was endangered, incentivising the Slovak government to request assistance in adopting an MPP (World Bank, 2002c). Similar to the previous cases, the main role of the Bank was to provide technical assistance in drafting and implementing MPP legislation (World Bank, 2002c). Slovakia implemented its MPP in 2005, but the project completion report from 2008 shows a more nuanced perspective: after its implementation, the successor government modified the structure of the MPP, making it optional for new insurees, thus repealing its mandatory nature (World Bank, 2008). In this context, the Bank rated the government’s performance as ‘unsatisfactory’ (World Bank, 2008). Two new aspects are visible in this case: first, for the first time, there is a more defined indirect link to the EU and, more specifically, the Eurozone. The documents mention fulfilling the Maastricht criteria regarding public sector expenditure as a reason for the project. Thus, while not directly involved in promoting MPPs, the EU’s liberal paradigm and the desire of CEE countries for European integration are probable contributing factors to MPP adoption. Secondly, deviations from the Bank’s vision of MPPs lead to at least a soft form of ‘naming and shaming’ by rating the performance of the respective government as unsatisfactory.

Finally, the case of Croatia is similar to the previous one, with the sole particularity that the Croatian government already started the process of pension pillarisation when the Pension System Investment Project emerged in 2002, and the language of the documents suggests strong support for the introduction of an MPP (World Bank, 2000a, 2002b). In this case, the main activity of the Bank is still focused on technical assistance, but it is unclear and rather improbable that its role was as important as in the other cases, as Croatia already adopted legislation preparing the introduction of an MPP in 1998 (World Bank, 2009). Unsurprisingly, as the process of pension pillarisation was already in an advanced stage, Croatia implemented the MPP in the same year as the World Bank project started.

These short comparisons of the six additional World Bank projects reconfirm the main finding in the case of Romania: the World Bank is usually a vital factor in the adoption and implementation of an MPP, mainly through its technical assistance. However, the cases of Lithuania and Slovakia especially show how the World Bank is also a political actor and not just a mere technical assistant. The Bank does indeed use its influence to create the framework for an MPP, even when there is not enough political will, and uses its reports to ‘name and shame’ countries that deviated from its vision of pension pillarisation.

6. Discussion – Surely Politics, Maybe Demographics

The quantitative and qualitative analyses have shown different insights into the role of political determinants for MPP adoption. While each part on its own provides clear information, I further discuss and compare the results together to provide a more in-depth understanding of the topic, also considering the more theoretical aspects presented in the theory section.

Government ideology does seem to play a role in the overall adoption of MPP. While some quantitative models presented insignificant results, the overall results and the qualitative case study show a major difference regarding how parties with different ideologies relate to MPPs. However, this does not explain why both left- and right-wing governments would opt for pension pillarisation in some circumstances. The extensive theoretical literature has shown that there are fundamental differences between the policy options of parties with different ideologies based on the socioeconomic group they represent.

One possible explanation could be the bargaining results that create governmental coalitions. If a government is formed through a coalition that is not entirely right-wing or left-wing, then policy programmes would normally be negotiated during the coalition-building phase. In that case, right-wing or left-wing parties could choose to concede their position on pension pillarisation in exchange for achieving the policies they favour in other areas, especially taking into account that pensions are a sensitive topic that affects the whole population and can cause a popular backlash.

This, however, does not apply to the case of Romania, where a fully left-wing government opted and implemented the first steps for pillarisation. Throughout all documents analysed, the topic of creating a favourable context for EU accession was present. All adopting countries other than Sweden are from CEE and were not part of the EU, but candidate states when adopting, except for Lithuania. Lithuania, in particular, was aiming to fulfil the Maastricht criteria for the Eurozone at the time of MPP adoption. There could be regional specifics, especially considering the similitude of their transition from communism to a free-market economy and their process of EU accession. The wish to show more liberal policies than the EU (Orenstein, 2008a) and prove the worthiness of accession can be a plausible alternative explanation.

On the other hand, there are clear insights that show that the World Bank played a significant role in MPP adoption, even though some statistical models were not significant. Even in the case of Romania, it is unclear if and how the bank influenced the government to opt for pension pillarisation, but there will be clear World Bank involvement once the decision is made. Returning to the model of transnational actor influence, the case study shows the last two stages of the policy process: diffusion and implementation. The Bank used its experience to assist other countries in helping Romania design its regulatory framework and implement the system. It also provided loans through multiple projects to aid the country financially. Thus, it used both its soft and hard power, as outlined in the theoretical section. Its collaboration and strong connections with the government and bureaucracy are the empirical equivalent of the transnational coalition lobbying a domestic veto player in the theoretical model.

Comparing Romania to the other six countries with World Bank projects, especially with Lithuania and Slovakia, shows the mechanisms through which the World Bank exerts influence in countries lacking political support for pension pillarisation. The insistence on the introduction of an MPP, especially in Lithuania, seemed unsuccessful at first but ultimately did result in the adoption of an MPP. Naming and shaming countries in the implementation reports by labelling the performance of the Slovak government as unsatisfactory, for example, is another mechanism through which the Bank uses its power to constrain government decisions. These actions follow the model of transnational actor influence, but more importantly, they are effective. There is no country in the sample with a World Bank pension-related project that did not adopt an MPP. While at least a minimal level of political will for pension reform seems to be a necessary condition, the presence of a World Bank project appears to be a sufficient condition for MPP adoption.

One aspect the quantitative analysis did not take into account but could be important for the time after the financial crisis is the policy-learning and position-adjusting inside the World Bank itself. While it has not completely abandoned the multi-pillar model, the perspectives inside the bank shifted towards more focus on a strong public pillar (Heneghan, 2022).

All aspects considered, there is a clear link between the political ideology of governing parties, the involvement of the World Bank, and the adoption of MPP in both Romania and Europe. In addition to it, the most puzzling result from the control variables was the lack of robustness for demographic factors. The rationale of the Romanian government follows the Bank’s idea that a PAYG system is unsustainable in an ageing population. The quantitative results do not support the argument that public economics considerations such as this are the driving factor for MPP adoption. One might wonder if not demographics per se, but the perception of demographic decline in the eyes of decision-makers plays an important role. However, this is a topic for further research.

7. Conclusion

This paper has analysed the role of political factors in the decision to adopt an MPP. By employing a mixed methods perspective, I found that both the ideology of the governing party and the involvement of the World Bank are significant in the policy design and implementation process. While the results for these political variables were not always statistically significant, there is a puzzling insight regarding the role of demographic factors. The analysis makes clear that considerations such as ageing do not have a statistically significant role in the decision for or against an MPP.

There is the question of the role of this paper in the existing and future literature. This paper has filled a research gap by looking primarily at political factors through a quantitative perspective and provided insights that can open perspectives for further research. First, the unexpected results of political ideology in the case of Romania can be studied further. An approach that particularly looks at the transition from communism to capitalism and the European integration in CEE could provide additional insights, as MPP adoption is often a post-socialist phenomenon. Another important aspect was how the World Bank is involved. While I showed its influence in the policy design and implementation stages, the internal decision of a government to pursue the legislative creation of an MPP was not addressed. It would be beneficial to see what the Bank does at this stage to get a complete picture of its involvement. Demographics are an additional factor to analyse in-depth. One could analyse whether demographics actually render public pensions unsustainable or if it is rather an excuse for governments to pursue reforms that otherwise would receive popular backlash. All these possible research topics show that the topic of pension pillarisation and MPP has not been researched sufficiently. Pension reform remains a complex issue that needs to be further studied.

Returning to my research question, the influence of transnational actors is the main political factor that determines the adoption of an MPP. The puzzle has also outlined the broader question of whether pension reform is an economic or political decision. While economic considerations, like demographic dynamics, do not have a clear impact on such reforms, there is a clear political component to it. While all European countries face the problem of ageing to some extent, only some opted for an MPP. Without considering the role of the World Bank, the image would be incomplete. To conclude, political factors surely influence pension pillarisation.

Eduard-Alex Ciuhandu