Liberal democracies combine core ‘macro-institutions’ (like free elections and control by legislatures) with swarms of supportive ‘micro-institutions’. By contrast, semi-democracies keep only the façade of macro-institutions, subverting a range of critical micro-institutions so as to make political competition and popular control a hollow sham. Drawing on a new book, Patrick Dunleavy explains why these developments mean that political science has to get a lot more granular and sophisticated, instead of focusing just on ‘toy models’.

Institutionalism has been a big movement in political science over the last two decades. It’s an approach that argues for the importance of a country’s institutions and their evolution in shaping its political development, as opposed to alternative causes – such as sociological or cultural influences, or political economy changes, or the abstract working of ‘rational man’ imperatives, to name but three. Yet in recent years the advances made by institutionalist analysis have seemed to taper off, and I argue here that this reflects an over-focusing on relatively few macro-institutions that form the standard repertoire of political science.

Most research about the pre-conditions for and influences shaping liberal democracies’ development still focuses on some tens or dozens of macro-institution variables – such as the kind of electoral system being used, the number of parties in the party system, the level of ‘consensus’ in legislatures or executive government or the fiscal decentralisation of government. Much modern research is still just about trying to quantify macro-institution variables’ effects more precisely (with more statistical controls), or to understand their operation in more qualitative ways. But a relatively small causal repertoire is still being discussed, in a ‘big data’ era when we might expect more developed explanations to have become a lot more feasible.

Lessons from bio-genetics

Past history offers many examples where social and political scientists have been influenced by developments in the STEMM disciplines (science, technology, engineering, mathematics and medicine) in how they seek to understand society. Of course, no direct read-across can occur – but the methods involved in STEMM research often inspire social scientists to do something similar, if they can. For instance, in a range of areas now, ‘big data’ and the application of artificial intelligence are likely to have extensive consequences for social science methods, just as they already have in STEMM and business research. And the models that STEMM scientists develop often furnish influential analogies – especially in understanding how complex causation of events can work.

In terms of causation analogies, the modern development of genetics research has been most recently influential. A decade or more ago geneticists confidently anticipated that they would be able to ‘explain’ the onset of many different human conditions and diseases by identifying small numbers (ten to a dozen) of genetic markers in the human genome – and that this in turn would open the way to potential remedies at the genetic level. The first part of these expectations has been confounded however by the far greater complexity of genetic conditioning than anticipated. The modern picture is that:

Many small genetic changes are involved in the expression of a single trait, and each change is correlated with a tiny tweak to the human form. To find the tiny effects that individual letters of the genome have on traits, disease, and behaviour, you need enormous data sets to separate signal from noise.

In particular, although there are some critically important ‘main effect’ genes, how they operate turns out to be fundamentally shaped or conditioned by many other ‘small effects’ genes that often switch on or off, or radically modify, the impacts of the ‘main’ genes. The result is a far more complex and holistically shaped set of influences, requiring the most careful analysis (and data sets of millions of cases) to unpick hundreds of different effects operating simultaneously.

Political analysis has neglected micro-institutions

This rapidly improving picture is interesting when set against the far simpler and dreadfully static range of causal patterns that are still being explored in political science, economics and sociology. The approach we adopted in conducting our 2018 democratic audit of the United Kingdom is informed by a different approach, one that assigns a lot of significance to multiple factors interacting in highly complex causal nets.

To start with, creating and maintaining any state is a not a simple thing. And controlling that apparatus in liberal democratic ways greatly increases that complexity. It involves meeting many different necessary conditions, all at the same time. These inescapable linkages justify the approach adopted in the 2018 Audit, of making an in-depth assessment of the quality of the UK’s democratic life across multiple different topic areas.

If the rise of semi-democracies has taught us anything it is that a genuinely democratic polity is constructed both from a small set of macro-institutions (such as a voting system, or a Parliament), plus dozens (perhaps hundreds) of micro-institutions. Micro-institutions are small-scale rules and regulations, or minor cultural practices. They often sit well outside the scope of any formal ‘constitution’, instead lurking in the detailed supplementary practices or mores that grow up around how macro-institutions operate. They are also often found in administrative or legal codes that apparently have little direct connection with the macro-institution they shape. Micro-institutions often play complex roles, some switching on or off the effects of macro-institutions, and others changing radically how macro-institutions operate.

For example, an electoral system is a macro-institution, but its key supports are the sets of micro-rules governing which parties or candidates can stand for elections, or how constituencies are drawn, or who can vote, or how politically balanced any state-controlled media must be between parties). In semi-democracies (like Russia) elections are still held and votes counted as in mixed electoral systems in the West, but parties and candidates opposing Putin are hounded out of existence by administrative regulations, and media coverage is highly biased. Even within Western countries, an electoral system like Spain’s is apparently a list Proportional Representation one, except that seats are so mal-distributed in relation to population (in ways favouring rural provinces, and disadvantaging big cities) that every election shows a very high deviation from proportionality.

Similarly, the USA’s worrying recent slide down the democracy rankings reflects the increasingly unbridled activities of partisan state legislatures in gerrymandering constituencies and (in Republican states) engaging in a plethora of ‘voter suppression’ tactics, like striking from electoral rolls anyone who has not voted across two elections. Even in the core of the USA’s designed constitution micro-institutions matter a lot. For example, James Maddison designed the US Electoral College as an elite-level safeguard for ensuring that only ‘moderate candidates’ would reach the Presidency – but the subsequent development of strong parties quickly reduced the College to a constitutional cipher. Today, a president with a strong activist base allowing them to control their party, and with that party having ‘undivided government’ dominance in Congress, can evade almost all checks and balances in the US constitution, at least for a time. By skewing judicial appointments to the Supreme Court a ‘lucky’ president may even set the tone of federal government for decades – as Trump’s record amply shows.

Another clear example of micro-institutions’ enduring importance concerns the UK’s Parliament’s role in budgeting. Since the English civil war in the seventeenth century was resolved by restoring the monarchy in 1659, our (uncodified) constitutional law says beyond any doubt that the House of Commons sets the government budget. But a tiny little rule, sitting in the Standing Orders of the House for decade after decade, also says that no MP can present any proposal for spending even £1 of public money unless they have a certificate signed by a minister, which is never given. At a stroke this requirement means that only ministers can present a budget, and that Parliament can perhaps cut spending out of it (although it has not done so in modern times), but can never add in anything new.

This is a key foundation for the normal de facto dominance of the government over the House of Commons, no matter what the formal or apparent constitution may say. The rule privileging only ministerial proposals for spending has also migrated to many other Anglo-influenced parliaments (like Australia, New Zealand, Canada, Ireland, and even the French fifth republic). In principle, of course, a simple majority of MPs could amend the Standing Orders to remove the budget certificate requirement – but the cultural and attitudinal rethink needed for any such change after so long means that it is not something that ever ‘comes up’. And of course, the elite of the top two parties have a joint incentive to keep the status quo in being. Following a more one-party logic, in a flagrant attempt to rig political futures, in 2015 the Conservatives used their (temporary) majority to push through an ‘English votes for English laws’ restructuring of Commons decision-making with no public consultation or legal testing, using Commons standing orders. This has so far proved ineffective, since the 2017 election brought another hung Parliament.

‘Toy models’ in political analysis

We have only just begun to absorb the importance of micro-institutions, so that many questions around them are up for discussion – such as how to distinguish one, and (quite importantly) count how many there are. Systematically mapping micro-institutions is just beginning, but the relevant numbers within the UK polity are likely to be numerous – on theory grounds alone. The implication of micro-institutions is that many more combinations of ‘big’ and ‘small’ institutional arrangements matter than either quantitative analyses (still testing ‘toy models’) or institutional theorists are prepared to admit.

How many combinations might matter in real-life situations though? Suppose that there are three institutions that operate as switches with a range of settings, running in 1% increments from 0% (fully off) to 100% (fully on) for each switch. There would then be 833 different combinations of switch outcomes. Extend this scenario to ten such switches acting at the same time and the number of combinations exceeds 2 million combinations.

If these set-ups seems unlikely consider that in 2010 in a House of Commons with eight parties there were only three ‘minimum winning coalitions’ (those with no ‘spare’ members), of which only the Conservative–Liberal Democrat alliance was judged feasible by elites – because Conservative-Labour co-operation was deemed unthinkable, and a multiple-party non-Tory coalition was also ruled out. In 2017 hung Parliaments returned, but this time no minimum winning coalition passed the party elites’ acceptability tests. Instead the simplest one-party minority government formed, with DUP ‘confidence and supply’ support. Theresa May’s subsequent efforts to conduct the Brexit negotiations by seeking only an incumbent -party solution subsequently lapsed into delays and apparent confusion because there was no consensus within the Conservative party (and also with the DUP) on how Brexit was to be operationalized.

Some scientific and prescriptive implications

The importance of dozens of micro-institutions means that political scientists need to be far more modest than most journal articles in the discipline’s top journals currently claim to be. We need to admit that as yet we have only some ‘broad brush’ ideas of how a few macro-variables interact to sustain liberal democracy or not. And we have barely begun to scratch the surface of assessing micro-institutions’ significance – especially in switching on or off, or altering, how macro-variables operate.

At present this means that we are limited to undertaking a qualitative analyses of complex situations – although these can be organized systematically. However, there is plenty of room for bigger-scaled quantitative analyses (that are replicable and take full account of multiple variables) to add to and develop our over-arching understanding. So a call to study micro-institutions is not in any sense just another plea for continuing or prioritizing qualitative analysis.

Turning to the prescriptive implications for sustaining democratic advances, across all the chapters of the Democratic Audit 2018 we found that many different micro-institutions matter for the democratic quality of the UK’s political life. And this has some powerful prescriptive or normative implications for established liberal democracies. Political elites and citizens should take very seriously alterations or deteriorations in how the ‘small stuff’ of democratic practices are set up. As Democratic Audit’s co-founder David Beetham argued (p. 568-9):

An [underlying] assumption we have made in our work is of an inertial tendency inherent in social and political systems towards oligarchy and inequality, unless it is being actively resisted. This means that the work of democratisation is never finished and that established democracies are as much in need of critical assessment as developing ones.

To best sustain liberal democracy, we need a whole ‘swarm’ of micro-institutions to operate in effective ways to support the core of macro-institutions – and we should not tolerate persistent small defects that corrode or undermine overall democratic quality.


Note: the above draws on the The UK’s Changing Democracy: The 2018 Democratic Audit, published by LSE Press.

About the Author

Patrick Dunleavy is Professor of Political Science and Public Policy at the London School of Economics, and Centenary Professor at the University of Canberra. He is the lead editor of The UK’S Changing Democracy (LSE Press, 2018), published free and open access.


Print Friendly