“The Three Worlds of Welfare Capitalism”, part 2: Stratification, Pensions, and Crystallization

This article will round out the first part of the Esping-Andersen’s book from 1990. You can find the previous part here. Enjoy!

Chapter 3: The Welfare State as a System of Stratification

Many sweeping generalizations about welfare are incorrect. For example, you might hear someone argue that it lessens class divisions. But not all welfare is created equal: some welfare policies lessen class divides, but some enhance them. And the redistributive effect of even the same amount of welfare spending can vary greatly. Mirroring the regime types, there are 3 “ideal types” of welfare stratification.


The origins of conservative stratification comes from two related phenomena: etatism and corporatism. Etatism, pioneered by Bismarck, means giving preferential treatment to public sector workers to ensure their loyalty to the state. Meanwhile, corporatism ties benefits to where you work, harkening back to the medieval era. The pre-capitalist idea of “noblesse oblige” mixes with Catholic ideology to create a system that generally helps the poor more than the liberal type does. Benefits can also be occupationally-segmented: Italy, for example, has 120 occupational pension schemes.


On principle, liberalism is opposed to stratification, viewing state interference as creating paternalism and elitism. Interference is thus kept to a minimum through stingy, means-tested benefits. But the state is not a neutral actor in liberal regimes: rather, it’s actively involved in centering the market in society. This can be seen in schemes like tax deductions for occupational benefits: a liberal state artificially promotes such things over social democratic alternatives. This all adds up to a highly stratified system, with the poor relying on stigmatized means-tested benefits, the middle class on subsidized and individualized social insurance, and the rich on the market. “Liberalism’s universalist ideas were contradicted by the dualism and social stigma it promoted in practice.”


Unsurprisingly, the social democratic or socialist approach aims to build solidarity and prevent the divisions of state paternalism, exclusionary corporatism, and the market itself. Yes, socialists opposed the pacifying welfare of Bismarck for example. But welfare could be used as a base to strengthen working class power. This idea was not new: Marx thought similarly about the British Factory Acts, for example. As mentioned previously, early attempts to build solidarity outside the state through cooperative and fraternal societies proved inefficient. Not only was there stratification between members and non-members, but worker-financed funds were always severely weakened during economic downturns due to unemployment. The unemployed were also a problem for the socialist project, as they undercut working class power (making strikes and high wage demands less effective), while not having much power themselves. We’ll return to this problem in the next article.

Comparative Dimensions of Welfare State Stratification

Now, we turn to some data on stratification. Corporatism is measured by the number of occupationally distinct public pension schemes, and etatism is measured by the percent of GDP spent on public employee pension. Universalism is measured by an average of coverage rates for sickness, unemployment, and pension benefits. Lastly, average benefit equality is measured by the ratio of minimum to maximum pension benefits. As you can see, divisions into welfare types are not perfectly clear cut: countries have a mix of different kinds of policy.

We also have countries clustered by the strengths of various welfare state logics.

Of particular note are two extreme cases. In the US, conservatism and socialism score a 0, while liberalism gets the highest possible score. Conversely, in Sweden, conservatism and liberalism score a zero, while socialism gets the maximum score. Note that a score of 0 for liberalism doesn’t mean that Sweden has, for example, absolutely no means-testing, just that it scores low on all categories associated with the liberal welfare state.

4. State and Market in the Formation of Pension Regimes

Special attention should be paid to generally the biggest welfare expenditure: pensions, including both old-age and disability pensions. Governments grant very different social rights here, varying from generous public pensions to schemes that push for occupational or individual pensions. So, the coverage of different types of pensions matters. But how much of retired folks’ income is actually from different pensions matters too: government granting you a social right to a pension doesn’t mean the quality of that pension is good.

Pensions were not always significant though: they played a marginal role until World War 2. After all, as late as 1870, most employment was agricultural outside the UK. Less commodification of labor meant less of a reliance on wages or old-age insurance. Instead, the elderly relied on family support and poor relief. For example, in the US in 1927, private charity was over 6 times greater than public welfare spending. With an aging population and further commodification, demands for pensions grew. Both public and private pension spending grew, crowding out pre-capitalist support and poor relief.

Restructuration in the Post-War Era

“The hope that a minimalist state and buoyant market would complement one another in a harmony of capitalism and welfare was falsified even before the Depression shattered it entirely.” In 1939, the countries we’re studying averaged pension coverage rates of 40%, with replacement rates of 15.5%. This was woefully inadequate, even if you were lucky enough to get a pension (only 5% did in the US). But much changed in World War 2. First, due to wage controls, employers started offering fringe benefits like pensions to attract workers. The high taxing and spending common to the war also expanded the baseline for what was acceptable for government to do. Some numbers from the US: from 1945-1950, private pension spending increased 68%. From 1950-1960, it increased a staggering 364%.

Still, many were left behind. Pensions often worked on an actuarial basis, meaning you got out what you paid in, plus returns on investment. This hurt mobile workers and women (who often took significant time off to raise kids). Worsening this were rules that prevented you from getting a pension unless you put enough years in.

Unsurprisingly, we can separate out 3 pension regimes.

1. Corporative: segmented based on status and occupation, which divides employees and glorifies management.

2. Residualist: market-dominated. Governments use tax expenditures to push private pensions (like 401k’s in the US), in a classic example of the state creating markets.

3. Universal: state-dominated. Grants population-wide social rights.

We can differentiate these by looking at pension spending by type, as below:

Chapter 5: Distribution Regimes in the Power Structure

Several explanations have attempted to explain why nations crystallize into certain welfare structures. First is functionalist moderation, which argues that economic development naturally leads to average folks demanding services the market can’t provide, which governments respond to. However, this can’t explain why similarly developed countries have different welfare state structures. There’s also the power-centered approach, which argues that governments are not necessarily responsive to the people. As such, policy is changed through power mobilization, of the working class. However, this is a very one-dimensional view: the working class can be mobilized in lots of different ways, depending on the structure of unions and the role of parties. Whether right-wing parties are unified and the structure of parliament may also play a role in how this power mobilization works.

It is at least clear that “being captive to market forces, workers are irrational if they do not seek a modicum of decommodification”. This protects them from the whims of bosses, but it also promotes solidarity, helping to stop workers from undercutting each other through strikebreaking, for example. Thus, “whether under the banner of social democracy, communism, or laborism, almost all parliamentarian labor movements converge with respect to their principles for social policy and designs for welfare-state reforms”. However, the working class may choose to organize towards ends other than class solidarity, like a focus on family and the Church.

We can’t limit ourselves to quantification of the modern situations in developed countries either. We must look towards history too. We’ll use a few variables to study how welfare states form over time. First is working-class mobilization, a weighted average of legislative and cabinet seats for leftist parties. This will be denoted by WCS (weighted cabinet shares). Seats are used instead of vote share to account for differences in how parliaments assign seats. Since union strength is highly correlated with this measure (with a coefficient of 0.816), and the focus is on welfare policy through legislation, we will exclude it as a separate variable. Next are variables measuring the strength of Catholicism, at first through measuring the share of Catholics in a country, and then through measuring legislative seats held by Christian Democrats, once those parties emerged around 1950. Third, we have a variable for absolutism, measured by the past strength and duration of absolutist rule, weighted by the year when universal suffrage was made law. We should expect this to correlate with etatism.

Lastly, we have two measures to account for non-power mobilization explanations. First is the average rate of GDP growth from 1960–1980 (when welfare states matured). Second is the rate of elderly folks relative to the population, since of course this would be a source of welfare expansion demands.

First, some notes on abbreviations: r is the correlation coefficient, B measures the slope of the relationship, R sq. measures how much variance in the dependent variable is explained by the independent variable, the higher F is, the less likely the relationship is from random chance, and N.S. stands for “not significant” here.
First, we simply measure the correlation between the variables above and social spending as a percentage of GDP.

Clearly, basically nothing has a statistically significant correlation. The exception is the rate of elderly in the population (indeed, once you control for this, the correlation with WCS in 1977 goes away). This is compatible with our hypothesis: different groups want different welfare structures more than a certain amount of welfare spending in and of itself. But of course, more elderly folks just means more people who need welfare, though it took time for them to become a significant force. It’s somewhat surprising that Catholic party strength or WCS is not even significantly correlated, though: we would expect their policies to generally increase spending.

The above table is noteworthy for the very strong relationship between corporatism and absolutism. This is a bit strange (we would expect it to be more strongly related to etatism), but still in line with our welfare typology.

Next is the dependent variable of “social security bias”, which refers to how much pension spending is public. Social democracy aims to crowd out private insurance in areas like this. We also add an independent variable measuring how long bureaucracy has been round: some theorize that, once established, welfare bureaucracies aim to increase their own power by increasing spending.

WCS explains ~47% of the variance here, lending some support to our theory of social democratic goals. Also, it’s important to note that the bureaucracy variable has an insignificant, but NEGATIVE relationship to social security bias, addressing a common right-wing talking point.

Lastly, we’ll see how well our measures correlate with the different welfare regime types.

The above data suggests that working-class power has not been successful at reversing conservative welfare structures. However, the opposite is true of liberal welfare. As expected, measures of conservative power are negatively related to socialist welfare structures, though not significantly.


The data overall supports our hypothesis: political differences have a strong effect on how welfare states are structured. But it also urges caution. None of these relationships are perfect: some are weaker than expected, some may be spurious. And history is always hard to put hard numbers to. We should also note that these relationships may not hold forever: whether they even hold today is something only more modern research can address. Plus, one important aspect has been largely left out of these discussions: how these welfare regimes approach employment. The next article begins to address that question.



Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store