Are the normative rationales of local autonomy related to democratic legitimacy justified? Evidence from 57 countries in Europe and beyond (1990–2020)

ABSTRACT

Frequently, increasing local autonomy is recommended to improve democratic legitimacy and government effectiveness. This paper attempts to verify to what extent these normative rationales are empirically sound. By using observations of local autonomy from the LAI 2.0 project (57 countries from 1990 to 2020) as independent variables and a set of governance indicators as dependent variables, we examine the possible impacts of local autonomy on various input-, output-, but also throughput-related aspects of legitimacy. The results show that local autonomy positively influences various aspects of input and throughput legitimacy, but not across the entire array of examined indicators. Our fixed effects regression models indicate that local autonomy is significantly related to the existence of elected local government, quality of democracy, accountability, transparency, equality of participation and openness. This research allows us to justify the normative claims of local autonomy facilitating efforts towards a more legitimate functioning of local governments.

KEYWORDS:

• Local autonomy
• democratic legitimacy
• local government
• decentralisation
• comparative research
• indext

Introduction

In recent decades, local autonomy and decentralisation have both been advocated by many important European and international institutions as a key tool of ‘good governance’ (UCLG 2008; UN Habitat 2009; White 2011). Decentralisation reforms transferring responsibilities, resources, and political decision-making towards intermediate or local government have silently been sweeping the globe since the 1980s (Ivanyna and Shah 2014; Rodriguez-Pose and Sandall 2008). Increasing the autonomy of the levels of government closer to the citizens is frequently advertised as a strong recommendation to enhance democratic legitimacy, government effectiveness, and governance processes. This improvement may happen through more involvement of citizens in the political process, an increase in the accountability of the decisions, the improvement of economic efficiency, the encouragement of healthy local competition, the support in policy experimentation, or the protection of macroeconomic and political stability (Andrews and De Vries 2007; Hankla 2009; Treisman 2007).

In the European institutional context, for example, in 1985, the Council of Europe adopted the European Charter of Local Self-Government, an international legal instrument ensuring the protection, evaluation and promotion of decentralisation and local autonomy principles, which entails the existence of local authorities endowed with democratically constituted decision-making bodies and possessing a wide degree of autonomy with regard to their responsibilities, the ways and means by which those responsibilities are exercised and the resources required for their fulfilment (Council of Europe 1985, preamble). Beyond the European context, the OECD provided guidelines to policymakers for decentralisation to produce its outcomes in terms of democracy, efficiency, accountability, regional and local development (OECD 2019).

Despite this widespread international trend towards decentralisation and local autonomy, it turns out that the normative values of local autonomy have not often been empirically tested in a comparative design in the literature, or by focusing only on a very specific expected consequence in a particular context. For example, Gendzwill (2021) shows that local autonomy had a positive effect on the local voter turnout, measures through the aggregate turnout rates in local elections across 21 European countries between 1990 and 2014. Ladner et al. (2019) do not focus in depth on one central component of the political system that autonomy could impact but on several. However, this research is also limited to the European context. Comparative studies on the consequences of decentralisation, for which local autonomy is expected to give a positive value are plenty (Page 1982), but they do not distinguish between subnational levels of government and do not take into account whether governments have a say in how tax revenues are used.

According to us, the lack of large-scale empirical studies on the expected consequences of local autonomy on different aspects of the political system comes from three main reasons. First, the normative claims on local autonomy and decentralisation are empirically seldom questioned, since their positive implications are sometimes counterbalanced by adverse impacts (Musgrave and Musgrave 1976; Prud’homme 1994). Second, the concepts of decentralisation, local autonomy, and democratic legitimacy very often get mixed up and there is a variety of divergent views and meanings that can be encompassed in these labels (Fleurke and Willemse 2004; Page 1982; Schmidt 2013). Third, comparative researchers in these fields are confronted with a lack of good comparable data at the local level due in large part to the wide variety of territorial models (Bouckaert and Kuhlmann 2016).

Against this framework, the aim of this article is to address the implications of local autonomy for various aspects of democratic legitimacy. On the one hand, we rely on a large set of external governance indicators as dependent variables. On the other hand, we take profit of the recent update of the Local Autonomy Index (LAI 2.0, see Ladner, Keuffer, and Bastianen 2023) to dispose of a multidimensional measure of decentralisation of local government on six continents as an independent variable. With the variables at our disposal covering 31 years from 1990 to 2020, we test the link between different types of government legitimacy and local autonomy. This perspective leads us to go beyond the underlying normative rationales of local autonomy in asking whether the politics of autonomous local governments ‘have an independent impact on the well-being of their citizens’ (Wolman, Goldsmith, and Newton 1992, 45). Indeed, by achieving their aims, local government reforms may affect the citizens’ day-to-day life. To follow this ‘freedom to’ approach (Pratchett 2004, 365), it is necessary to look more precisely at the causes of the different local government activities (Wolman 2008).

This paper is structured as follows: The first part discusses the various perspectives and normative rationales on local autonomy and local governments’ democratic legitimacy, as well as the link between these two concepts. The methodological part explains the construction and operationalisation of the main variables and present the quantitative approach. The results are then presented, which makes it possible to grasp the link between local autonomy and democratic legitimacy. Finally, the paper ends with a discussion of the main findings and some concluding remarks.

Theoretical considerations and hypotheses

The theoretical perspectives and normative rationales on local autonomy

Local autonomy has generally been apprehended in the literature as a relative concept and defined as the discretion local governments possess to act in ‘freedom from’ (Pratchett 2004, 363) the control by higher levels of governments. This perspective traditionally considers local autonomy in its vertical relationship with the state and is based on a constitutional or legal understanding of intergovernmental relations. There was a broad agreement in the literature that there was no valid theory of local autonomy, the concept being too diversified (Clark 1984; Kjellberg 1995; Mackenzie 1961). Firstly, this relates to the plurality of existing theoretical conceptions as well as to the evolution of the definition of local autonomy and its underlying values over time. Having been defined rather negatively as a right of democratically elected local authorities to resist imposed constraints, local autonomy has indeed seen its meaning broaden to include the positive capacity to realise local interests, as well as the means to implement other values (Chapman 2003; Hansen and Klausen 2002). Secondly, this comes from the interdisciplinary and multidimensional nature of the concept of local autonomy. The in-depth literature review conducted by Ladner et al. (2019) highlights that local autonomy is at the crossroads of different disciplines and conducts the authors to propose a theory-based and multidimensional concept. In this paper, we rely on the updated and extended Local Autonomy Index (LAI 2.0), as the LAI has brought a certain theoretical convergence regarding the core components of local autonomy, i.e., legal, functional, political, and organisational elements (Ladner, Keuffer, and Bastianen 2023).

The extent of local autonomy can be analysed through a formal approach to the distribution of tasks and competencies. It theoretically corresponds to the degree of decentralisation of a politico-administrative system: as local autonomy and decentralisation are treated as two sides of the same coin, a highly decentralised political system also means a lot of local autonomy and vice versa (Fleurke and Willemse 2004). This is more concretely true when there is only one intermediate level below the national state. This relates to a limit very often addressed to the publications based on fiscal decentralisation data, namely that the various subnational levels are not distinguished (Ebel and Yilmaz 2002; Vetter 2007). Moreover, these studies refrain from efforts related to taking into account other aspects of financial autonomy such as the origin of financial resources, the possibility of local governments to have a say in their use and the fiscal competencies of local governments (Akai 2013; Blöchliger and Kim 2016).

In this perspective, local autonomy normatively legitimises local government as a democratic institution as it implies an own sphere of authority to determinate the public services and local tax rates according to the preferences and needs of its community (Goldsmith 1995). In his review of liberal justifications of local government, Chandler (2008) argues that the normative rationale of local government is based on ethical and expediential grounds and that the second ones have gradually took precedence over the first ones over the course of the twentieth century. On the one hand, the ethics side relates to the intrinsic value of local government, as an extension of individual freedom and regardless of its value for the state as a whole. This argument was, for example, theorised by Mill (1966) who argued that the liberty of the individual to follow one’s own beliefs implies a corresponding liberty in any number of individuals to make decisions that regard them jointly. This implies that local governments adopt a democratic structure so that individuals can determine the services they want for the group that it possesses resources to provide these services, and that its integrity and activities are relatively preserved from higher levels of government (Chandler 2010). On the other hand, expediency justifies local government to the extent that it serves the purposes of higher levels of government. It focuses on the democratic and efficiency-related comparative advantages local governments possess compared to higher levels of government, since its closeness to citizens enables them to determine freely what matches predominantly their own needs and enables it to provide more tailored services. This implies the ‘efficiency-democracy’ dilemma (Goldsmith and Newton 1983; Pierre 1990).

Local autonomy can also be considered not as a positive value in itself, and in this sense, it is rather its real impacts and the underlying normative logic that come to the fore. In this ‘freedom to’ (Pratchett 2004, 364) perspective, local autonomy is defined as the ability of local governments to have an independent impact on the well-being of their citizens. Applying this analytical approach to compare local governments in the United States and the United Kingdom, the fundamental question posed by Wolman and Goldsmith (1992, 3) is as follows: ‘Do local governments in urban areas have autonomy in the sense that their presence and activities have independent impacts on anything important? Does urban politics matter?’ Accordingly, in this paper, we do not view local autonomy ‘through romantic eyes’ as it has been to a great extent the case in the dedicated literature (King and Stoker 1996, 24), but empirically test the effectiveness of the opportunities it offers. Only on such grounds we will be able to tell if it represents a feature that modern states aspire to.

Hypotheses of the consequences of local autonomy on democratic legitimacy

Local autonomy, defined as a policy space for local democracy (Ladner, Keuffer, and Baldersheim 2016), can be considered as a constitutive feature of elected local governments, in the sense that local governments do need an own sphere of power and discretion to cope with competing values, preferences and priorities and to provide local public services that are in line with the prevailing tastes and demands of the citizens. Due to their closeness to the population, it is accepted in normative terms that local governments are just as likely to respond to the preferences and collective needs of their citizens as to implement public policies locally (Pierre 1990). According to the expediential arguments exposed above, these functions of local governments can help to legitimise government as a whole.

As regards its function of bureaucratic efficiency, local government is a functionally efficient service-delivery agency which accurately translates public preferences into public policies (Sharpe 1970). For economists from the ‘public choice’ school, the role of local government is legitimised by the solution it offers to provide local public goods: ‘For a public good – the consumption of which is defined over geographical subsets of the total population [and for which there are not cost advantages to central provision] – it will always be more efficient (or at least as efficient) for local government to provide the [locally preferred] levels of output for their respective jurisdictions than for central government to provide any specified and uniform level of output across all jurisdictions’ (Oates 1972, 35).

Aside from normatively legitimising local government as being able to create and promote a well-functioning and attractive economic environment, local government can also be justified as a democratic institution, since it provides for a healthy territorial division of political power promoting stability. It may enhance local participation, bring forward representative, responsible and accountable local authorities, give opportunities for political skills to be developed and lead to a diffusion of power and promote inclusion within society (Jones and Stewart 1982; Tocqueville [1835] 1994).

Seen through the prism of systems theory, Easton (1953) was the first to distinguish between the input in the political system, i.e., the votes, demands or interests of citizens, and the output, i.e., the actual decisions and actions of governments. Drawing on institutionalism theory, Scharpf (1999) defines input-oriented legitimacy as taking into account the needs expressed by the people through their political participation and output-oriented legitimacy as the ability to govern effectively for the people. Vivien Schmidt (2013, 4) defines these two normative criteria as follows: ‘input legitimacy refers to the participatory quality of the process leading to laws and rules as ensured by the “majoritarian” institutions of electoral representation. Output legitimacy is instead concerned with the problem-solving quality of the laws and rules and has a range of institutional mechanisms to ensure it’. Using the European Union (EU) as an example, the author postulates that modern democratic theory lacks procedural governance practices, and proposes an additional legitimacy criterion, throughput: ‘Throughput legitimacy builds upon yet another term from systems theory, and is judged in terms of the efficacy, accountability and transparency of the EU’s governance processes along with their inclusiveness and openness to consultation with the people’ Schmidt (2013, 2). Consequently, the greater the autonomy enjoyed by local governments, the greater the latitude to decide on a broad spectrum of functions responding to issues raised by their citizens and, in so doing, the greater their responsiveness, accountability and transparency should be. In a similar way to input- and output-oriented legitimacy, local autonomy can thus be seen as the theoretical prerequisite for greater throughput-oriented legitimacy of local government. In this paper, we attempt to test these theoretical claims empirically. In doing so, the following general hypotheses can be made: the higher the level of local autonomy, the higher the quality of democracy (input-legitimacy), the efficiency of delivered outcomes (output-legitimacy) and the development of procedural governance practices (throughput-legitimacy).

Methodology

Operationalisation

In the following section, we detail the operationalisation of the independent variable and the dependent variables, as well as provide an explanation on the control variables. The choice of these variables relies on the theoretical and literature considerations discussed above.

As the main independent variable of local autonomy, we use the LAI 2.0, comprised of seven carefully assembled and weighted dimensions, all covering a wide variety of indicators related to the leeway of local governments in regard to their own freedom of local governance, but also in relation to higher levels of government. The overall index gathers the dimensions in one measure, to compare countries together more easily. The detail of the LAI, its construction and its theoretical and methodological justifications have been treated extensively in the literature (Ladner, Keuffer, and Bastianen 2021, 2023).

The dimensions of local autonomy are interesting components of the overall LAI. If a specific dimension stands out in the analyses, it is noteworthy to delve deeper into the details. The LAI distinguishes between the following seven dimensions of local autonomy (Ladner, Keuffer, and Baldersheim 2016, 2019): legal autonomy, access, policy scope, political discretion, financial autonomy, organisational autonomy and non-interference. In the results section, we highlight correlations between the various dimensions of the LAI to determine whether a specific dimension primes over the general index with regard to impact on legitimacy indicators. We then examine said impact on the same selected array of indicators measuring aspects of input-, output- and throughput legitimacy.

For our dependent variables, we operationalise the various indicators of legitimacy in the following ways. For input legitimacy, we use several measurements that all refer in a certain way to participatory quality, i.e., deciding on policies based on the involvement of citizens and their preferences (Scharpf 1999, in; Caby and Frehen 2021). Defined as such, we include measures of satisfaction with democracy, effective political participation whether institutional (turnout at elections) or not (e.g., petition, demonstration), trust in government, whether local government is elected or not, and finally, an overall measurement of the quality of democracy.

On the other side, output legitimacy can be measured by any policy outcome that performs in line with the interests and expectations of the concerned citizens (Scharpf 1999, in Caby and Frehen 2021). Output-wise, we integrate data on corruption as well as the proportional government debt-to-GDP ratio, and finally, in a more candid and unusual perspective of output legitimacy, a measurement of happiness. This latter indicator is well aligned with the concept of autonomy as the freedom of local governments to have an impact on the well-being of their citizens (Wolman and Goldsmith 1992, in; Pratchett 2004).

Throughput legitimacy, on a conceptual level, is comprised of the key aspects that are accountability, transparency, inclusiveness, and openness. Schmidt and Wood (2019), discussing how to operationalise the aforementioned constitutive elements of the ‘umbrella concept’ that is throughput legitimacy, suggest a few possibilities. According to them, throughput legitimacy offers distinctive normative criteria: accountability, transparency, inclusiveness and openness. They also acknowledge that commonalities with input- or output-aspects of legitimacy may exist, but that throughput legitimacy indicators are indispensable in multi-level governance arrangements (Schmidt and Wood 2019, 728), as is the local government landscape that we examine presently. Aspects of throughput legitimacy can be comprised into four categories (Schmidt and Wood 2019; Caby and Frehen 2021):

• Accountability: presence of a clear mandate, platform and procedural framework ensures effective and quality deliberation in decision-making, fostering open and honest discussions with citizens and stakeholders.

• Transparency: achieved when stakeholders can access clearly structured information and supporting materials used in the process.

• Inclusiveness: implies that diverse stakeholders from the pluralistic community are integrated, ensuring representation of all relevant groups.

• Openness: stakeholders should have a seat at the table, the opportunity to present their arguments and access to policymakers and their arenas, promoting an open and accessible decision-making environment.

There are also a certain number of control variables that we used throughout the various analyses. First, membership to the OECD is considered, as many countries are grouped together in this organisation which, as mentioned previously, produces a set of guidelines and recommendations towards more decentralisation and better governance outcomes. Briefly, two other memberships were considered as controls (EU and Council of Europe), but these are exclusively Eurocentric, and the goal of this paper is to provide general considerations that most widely cover countries across all parts of the globe. Second, mean size of local governments is controlled for, taking into account postulates regarding better responsiveness of the elected representatives in small municipalities (Dahl and Tufte 1973). This variable is calculated based on subnational measurements of the population and the number of local governments per subnational unit, sourced from the LAI 2.0. Finally, we have included a logarithmic scaled measure of GDP per capita to account for economic growth level, as countries with a higher GDP per capita tend to appear more often in the top tiers in most of the datasets on governance outcomes. We expect richer countries to score better on a variety of our chosen dependent variables. This is also corroborated in wider literature on the links between economic growth, democracy, and quality of governance (Cooray 2009, Rivera-Batiz 2002).

Data

Perhaps, the largest roadblock of this endeavour is the search for and compiling of adequate data sourced from secondary datasets, in a way that best represents the initially established operationalisation of variables, yet also covers the largest common ground, spatially and temporally, with the LAI data. One of the biggest limitations of the current research article is related to the availability of large-scale, comparative, longitudinal data, whether at the local level or aggregated higher level. The current selected variables and related data sources are the results of multiple iterations between theoretical considerations and available empirical evidence, ending in a pragmatic compromise that best fits this paper. A full overview of the operationalised variables and the data used for our analyses has been included in the supplementary material (see Appendix 1).

The independent variables data on local autonomy and its seven dimensions as well as almost all the control variables (except for GDP, which is taken from the IMF database, see Mbaye et al. 2018) are provided by the LAI 2.0 dataset (Ladner, Keuffer, and Bastianen 2021, 2023). For the dependent variables, a series of external datasets have been used as secondary material.

First, Christopher Claassen’s Democratic Satisfaction dataset covers almost 50 years (1973–2020) for 132 countries and is thus one of the more complete datasets at this level (Claassen and Magalhães 2022). The OECD (2023) Trust Survey dataset has been used to gather data on citizens’ level of trust in government. It covers the period from 2006 to 2022 and therefore does not align with the first half of the observations of the LAI (1990 to 2005). The Democracy Barometer offers an index that assesses the quality of democracy in 53 countries from 1990 to 2017 based on political rights, civil liberties, and the functioning of government. It provides both aggregate scores and individual indicators. We have used this dataset for our variables on effective political participation, transparency, inclusiveness and openness. Varieties of Democracy (V-Dem v12, see Coppedge et al. 2022) is a comprehensive dataset that provides a wide range of hundreds of indicators to assess the quality of democracy. It covers areas such as electoral processes, civil liberties, rule of law, and many more, spanning over multiple centuries and covering almost every country in the world. From this dataset, we extracted our relevant indicators on local government election, quality of democracy, free and fair elections, corruption and accountability. The World Happiness Report (Helliwell et al. 2021) is an annual publication that ranks countries based on their levels of happiness and well-being. It is produced by the Sustainable Development Solutions Network, an initiative of the United Nations. The report uses various indicators, including GDP per capita, social support, life expectancy, freedom to make life choices, generosity, and perceptions of corruption, to assess and compare the happiness levels of different countries.

The data for the throughput legitimacy variables has been selected as follows: for accountability, we have opted for V-Dem’s (Varieties of Democracy) accountability index, which is an aggregation combining vertical, horizontal, and diagonal accountability, respectively, referring to citizens’ ability to hold government accountable, checks and balances between institutions and oversight by civil society organisations and media (Coppedge et al. 2022). Transparency, inclusiveness, and openness measurements are taken from the Democracy Barometer (Engler et al. 2020), respectively, their ‘transparency’, ‘equality of participation’ and ‘openness of elections’ variables. Their transparency measurement comprises elements such as disclosure of party financing, absence of corruption, freedom of information and willingness for transparent communication. Equality of participation (used as our measurement of inclusiveness) takes components such as suffrage requirements and disqualifying criteria, percentage of registered voters as part of the voting age population, non-selectivity of participation whether electoral or alternative (including e.g., petition signings, attending lawful demonstrations, and more). Finally, openness of elections is probably the most restrictive measurement in this range of throughput-related variables when it comes to capturing the full definition of the concept, but it is used for lack of broader available data. This indicator includes measures on low legal hurdles for entry into electoral competition, effective contestation, and access to resources. It is therefore an important indicator of openness towards obtaining a seat at the table, albeit limiting its evaluations to institutional and electoral mechanisms of political representation, and not in a broader procedural sense.

After a series of statistical tests, some of the dependent variables were removed from the final analysis. This is mostly due to the lack of reliable and complete external data for said variables covering sufficient similar countries and years as the LAI. Various tests indicated irregular panel data and distributions on the dependent variables side, which is explainable because many of these datasets either do not present similar complete geographical and chronological coverage to the LAI 2.0 dataset (for full details on countries and years included, see Ladner, Keuffer, and Bastianen 2023). For example, this is the case for the variable on satisfaction with democracy, or the OECD trust in government data. After careful testing and selection, Appendix 2 provides descriptive statistics for all independent and dependent variables under scrutiny (see supplementary material).

Methods

Given the nature of the independent variable and dependent variable data, involving 57 countries over a 31-year period (1990–2020), to test the linkages over time across all considered countries, the data has been set up as a cross-sectional time-series.

Since most of the countries covered by the LAI 2.0 are OECD members, it is necessary to erase any concern on multicollinearity before proceeding further. During checks of independent and control variables, the VIF (Variance Inflation Factor) amounts to values between 1 and 1.3, well below the acceptable threshold of 5. This confirms that no remaining multicollinearity is introduced into our regressions. Pooling tests conclude that pooling data is unstable, and therefore OLS regressions are not employed during further steps. A series of Lagrange Multiplier tests (Honda, Breusch-Pagan) are executed to determine the presence of individual (i.e., country) or time effects. Here, we observe significant effects of differences across countries.

Finally, after a series of Hausman tests to determine whether a random effects or fixed effects model should be used to account for variation in the group of 57 countries under scrutiny, the choice of fixed-effects model usage come out as significant. After these procedures and the completeness of cases, we are able to investigate quality of democracy (covering input-aspects of legitimacy), corruption for the output-side of legitimacy, and accountability, transparency, inclusiveness (proxied by equality of participation) and openness, which could be kept for the analysis of throughput-legitimacy aspects.

Results

Before delving deeper into the analysis models, we start this results section by looking at how all the seven underlying dimensions of local autonomy correlate with our initially selected array of indicators measuring aspects of input-, output- and throughput-legitimacy. Table 1 shows a multitude of dimensions correlating with the various legitimacy indicators across the board.

Table 1. Correlations between dimensions of local autonomy, the overall index and aspects of input, output, and throughput legitimacy.

	Input legitimacy		Output legitimacy		Throughput legitimacy
	Local government elected	Quality of democracy	Corruption	Happiness	Transparency	Accountability	Equal participation	Openness in elections
LAI (overall index)	0.306**	0.475**	−0.493**	0.336**	0.325**	0.590**	0.190**	0.204**
Legal autonomy	0.235**	0.038	0.092**	−0.255**	−0.248**	0.051*	−0.052	0.262**
Access	0.158**	0.370**	−0.405**	0.017	0.070*	0.424**	0.083**	0.123**
Policy scope	0.002	0.093**	−0.235**	0.156**	0.348**	0.171**	0.066*	0.121**
Political discretion	0.155**	0.213**	−0.251**	0.130*	0.388**	0.340**	0.151**	0.242**
Financial Autonomy	0.140**	0.492**	−0.561**	0.547**	0.224**	0.569**	0.122**	0.019
Organisational autonomy	0.373**	0.338**	−0.252**	0.282**	0.253**	0.397**	0.272**	0.145**
Non interference	0.358**	0.337**	−0.248**	0.015	0.016	0.383**	0.109**	0.011
N	1657	1657	1657	385	866	1657	1208	1234

p-value thresholds as (p<0.01 followed by p<0.05)

A general broad outlook on this correlation table indicates that most dependent variables correlate with the core dimensions of political discretion, financial autonomy and organisational autonomy. Here, it is important to mention that the dependent variable measuring whether local government is elected remains a relevant choice, even though it might seem concerning that it appears to be closely related to a component of organisational autonomy. There are many differences between the measurement scales of the LAI 2.0 and the V-Dem 12 datasets, and most importantly, the former assesses the de jure situation, whereas the latter measures the de facto reality. The overall index is the strongest and most significantly correlating variable, which is why further analysis in this paper is focused on the overall index score as the independent variable. It also confirms the literature on local autonomy as a reliable indicator, underlining that all components of local autonomy are crucial in its appreciation of relevant causes and impacts (Harguindéguy, Cole, and Pasquier 2021; Ladner et al. 2019, 2023). Another main motivation behind this choice of focus on the overall index is that it reaches beyond most studies on local autonomy, which are often limited to observing allocation of financial resources (Akai 2013; Blöchliger and Kim 2016; Page 1982).

It is noteworthy to mention here that the number of observations (N) differs depending on where the dependent variable data was sourced from. For example, the happiness variable from the World Happiness Report only includes a limited number of countries from 2013 onwards. This should again be kept in mind during later analytical advancements. The lack of high-quality comparable data at the local level, as has been noted by other comparative researchers in this field (Bouckaert and Kuhlmann 2016), remains a persistent issue.

Pursuing with the fixed effects regression models, an overview of the results is presented in two tables: Table 2 for the regression models whose dependent variables are related to input- and output-aspects of legitimacy and Table 3 for the models covering four categories of throughput legitimacy. The latter dimension of democratic legitimacy is one of the major focuses of this article, since it has not been treated previously in the literature in the context of local autonomy, hence its separation into a different dedicated table, with the same number of tested variables as in the other one.

Table 2. Local autonomy, with controls and aspects of input- and output-legitimacy.

	Local government elected	Quality of democracy	Corruption	Happiness
Local autonomy	0.005**	0.005**	−0.0003	0.002
	(0.002)	(0.0004)	(0.0004)	(0.006)
OECD membership	0.082	0.022*	−0.009	0.297**
	(0.042)	(0.011)	(0.010)	(0.099)
Mean size of local government	0.337**	0.028**	−0.035**	−0.260**
	(0.017)	(0.004)	(0.004)	(0.064)
GDP per capita	−0.015	0.004	−0.025**	0.812**
	(0.017)	(0.004)	(0.004)	(0.098)
Observations	1,609	1,609	1,609	385
R^2	0.205	0.137	0.107	0.221
Adjusted R^2	0.175	0.105	0.074	0.083
F Statistic	100.062** (df = 4; 1550)	61.421** (df = 4; 1550)	46.499** (df = 4; 1550)	23.163** (df = 4; 326)

*p<0.05; **p<0.01.

Table 3. Local autonomy, with controls and aspects of throughput legitimacy.

	Accountability	Transparency	Equality of participation	Openness of elections
Local autonomy	0.011**	0.009**	0.009**	0.015**
	(0.001)	(0.003)	(0.002)	(0.002)
OECD membership	0.101**	0.055	0.029	−0.107*
	(0.023)	(0.053)	(0.067)	(0.044)
Mean size of local government	0.061**	−0.084*	−0.051	0.001
	(0.010)	(0.036)	(0.035)	(0.023)
GDP per capita	−0.014	0.306**	−0.032	0.136**
	(0.009)	(0.023)	(0.025)	(0.016)
Observations	1,609	862	1,190	1,199
R^2	0.125	0.260	0.015	0.226
Adjusted R^2	0.092	0.223	−0.028	0.193
F Statistic	55.241** (df = 4; 1550)	72.201** (df = 4; 820)	4.198** (df = 4; 1140)	83.782** (df = 4; 1149)

*p<0.05; **p<0.01.

Table 2 shows that local autonomy, in our designs, matters mostly when paired with input-related aspects of legitimacy. In more autonomous municipalities, the political system is more likely configured to have direct election of local government. Also, it is significant in increasing the overall quality of democracy, especially when it comes to clean elections, based on our available data. More autonomy provides more capacity to fulfill local interests, which in turn increases input legitimacy. This already confirms a major part of the normative rationales that have been theorised in the literature on local autonomy (Chapman 2003; Hansen and Klausen 2002).

In the case of output-related aspects of legitimacy, local autonomy does not significantly influence said outcomes. In our research, local autonomy does not significantly relate to less corruption or more happiness. Outputs of political systems are complex phenomena which are influenced by a plethora of factors that cannot be reduced to institutional explanations, such as local autonomy. Many personal and economic factors would need to be examined to find more satisfying explanations. Regarding economic factors, we do see that GDP per capita does account for some decrease in corruption and an important increase in happiness. Regarding the latter, the tinier number of observations due to less available data from the World Happiness Report introduces some bias in the interpretation of the fourth model examined here, and local autonomy could still be influential. The observed adjusted R2 values indicate a higher explanatory value for the models pertaining to input-aspects of legitimacy. This exacerbates the need for further examination of output-legitimacy aspects using other explanatory factors beyond our current focus.

Membership to the OECD also has some influence, but of lesser importance. This is partially explained by a certain degree of multicollinearity, which we deemed tolerable after the VIF test previously applied. For happiness, this is probably related to a certain amount of commonality between countries part of the OECD and countries sharing the highest GDP per capita, which we already determined to be a strong significant factor for this specific dependent variable.

Finally, still in Table 2, mean size of local government appears to have significant influence across the board, albeit sometimes in somewhat surprising ways. An increase in mean size of local government is related to a higher probability of local governments being elected (quite strongly so), an increase in cleaner elections and quality of democracy, less corruption, and a lower score in the happiness index. Specifically, the last result would make a case for the ‘small is beautiful’ thesis (Denters et al. 2014), although many other factors certainly come into play, and one should continue to keep in mind the lower number of observations regarding the happiness-dependent variable.

Table 3 presents our analysis on four dependent variables encompassing four main aspects of throughput legitimacy. Local autonomy has a noticeable and significant impact on all components of throughput legitimacy that were examined here. On transparency more specifically, our findings confirm, on a larger comparative scale, a positive influence of local autonomy on transparency practices (Keuffer and Mabillard 2020). In alignment with the ‘freedom to’ approach (Pratchett 2004), here we see that, to a certain extent, local autonomy offers opportunities for more effective procedures of accountability, transparency, and openness. It therefore enlarges the local institutional space for citizens, not only to better express their preferences and needs (input legitimacy, see Scharpf 1999), which in turn leads to better diffusion of power among other governance outcomes (output legitimacy, see Jones and Stewart 1982), but also to improved procedural governance (throughput legitimacy, see Schmidt 2013).

The model combining local autonomy and equality of participation (used to measure a segment of inclusiveness) has no explanatory value due to the adjusted R2 appearing negative. In this case, other explanatory variables need to be explored, which reaches beyond the scope of this article.

Going through the various controls we have applied to the throughput legitimacy models, we observe a certain impact of OECD membership on increased accountability, and conversely less openness. When left out, OECD membership does not interfere with the estimation, i.e., the direction or significance of the relationship between local autonomy and aspects of input-, output- and throughput-legitimacy. However, based on our theoretical considerations, it is important to include as a control, especially since it does show some influence, which is expected. Regarding mean size, larger local governments would lead to more accountability, but also less transparency. The latter relationship could possibly be attached to a proximity argument, where larger local governments would generate more distance between the population and their representatives. The level of GDP per capita, in the case of throughput legitimacy, seems to have lesser impact than on input- and output-aspects of legitimacy, here only exerting significant influence on measurements of transparency and openness. Overall, within the bounds of our research design, we can reasonably claim that local autonomy positively influences throughput legitimacy in increasing accountability, transparency, and openness.

Looking back at all our results, two out of our three main hypotheses are validated, these being whether local autonomy positively impacts elements of input- and throughput-legitimacy. However, our data and analysis does not allow us to validate the hypothesis on improved output-legitimacy. To reprise some of our introductory claims, increased local autonomy can indeed empirically enhance democratic legitimacy and governance processes, but our data leaves uncertainty when it pertains to government effectiveness. This also constitutes a valuable contribution to the literature on decentralisation and local autonomy, which rarely questions the latter’s positive values. As we have seen here, for example, regarding happiness, lacking and narrow data is still restraining us from confirming the general normative claim that local autonomy has a beneficial impact on citizens’ well-being (Wolman and Goldsmith 1992, 45). Notwithstanding the mixed results, our research still provides a substantial addition to the literature on local autonomy by showing, at a large-scale comparative level, its positive consequences on a wide range of aspects of well-functioning political systems.

Discussion and conclusion

In this statistical analysis of 57 countries over the period of 1990–2020, local autonomy positively influences aspects of input- and throughput-legitimacy. In the case of output-related aspects of legitimacy, local autonomy does not have a significant influence, since outputs of political systems are complex phenomena which are influenced by a plethora of factors. This means that two of our three outlined hypotheses can be validated. While we cannot be certain of its generalisability outside of these 57 countries, we believe with further research into the LAI 2.0 and with more detailed broad comparative data, we could highlight more specific intricacies pertaining to these relationships.

First, to answer our main question, our evidence leads us to affirm that the normative rationales of local autonomy related to democratic legitimacy are justifiable. Our findings corroborate the normative claims made on behalf of various international organisations that promote decentralisation efforts. For the practitioner community, this implies that it validates ongoing efforts towards decentralisation and building stronger municipalities, as many collaborative networks between local government experts and supranational structures have been pursuing. It also warrants continued monitoring of the implementation of the European Charter of Local Self-Government for the countries who have ratified the said document. It also reinforces recommendations by the OECD and the EU to improve and strengthen municipalities. Our work creates a path for more comparatively designed empirical studies on the value of local autonomy in improving legitimacy and good governance outcomes at the local level, and even higher levels of government.

Our research also goes beyond many of the theoretical elements treated in the body of the literature that we reviewed above. The current study shows that the expediential grounds on the importance of strong local government have indeed taken the forefront, as improvements at the local level also serve higher levels of government (Chandler 2008). Indeed, as a certain proportion of our dependent variables relate to higher levels of government and are also significantly influenced by local autonomy, we can confirm this theoretical basis. It also proves that local autonomy has an important and beneficial impact on citizens’ lives, as we mentioned in our introductory remarks.

The specifically detailed focus on aspects of throughput legitimacy, and the positive results related to them confirm the necessity to include throughput-related aspects of legitimacy in the general discussion on democratic legitimacy, as has notably been advocated by Schmidt (2013) and Caby and Frehen (2021). Good governance is not only attained by improving the input and output of politico-administrative systems but procedural concerns must also be addressed. The theoretical and practical values of throughput legitimacy are also validated in our results by the observation that they are less prone to other institutional or economic factors. This also reinforces recommendations towards better procedural governance, which fall beyond the realm of input- and output-aspects of legitimacy.

In a conscious effort to acknowledge some of the limits highlighted in our article, a majority of our sample of countries either belongs to the EU, CoE, or OECD, or a combination of said affiliations. An expansion of countries assessed by the LAI 2.0 would be warmly welcomed by the academic and practitioner communities. More specifically, in the framework of the issues and interrogations encountered in this paper, an assessment of local autonomy in more states that do not belong to any of these (or other) international organisations could determine to what extent the various arguments of membership and legitimacy just discussed are confirmable on an empirical basis. Various efforts (see for example Sundström’s ‘Local Quality of Government’ dataset) are propelled towards this direction. For now, we can affirm that local autonomy, in a decentralising democratic context, can provide the necessary room for more legitimacy.

Acknowledgments

We would like to heartfully thank Alyssa M. Taylor and Moulay Lablih for their technical assistance in the analytical process of our article.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/03003930.2024.2378128