Means tested programs are designed to steal from the middle class. They end up paying taxes and not getting benefits. Universal programs are better because the difference between what you pay and what you get tends to be lower.
Example
Means Tested for middle class:
Taxes = 80
Benefits = 0
Net = -80
Universal for middle class:
Taxes = 100
Benefits = 70
Net = -30
And I'm being generous there. Perhaps taxes are the same in those scenarios.
It's a libertarian fantasy that cutting universal benefits will lead to dramatically lower taxes. Governments maximize tax revenue, any reduced spending burden on universal benefits will get re-directed to welfare (for the poor, public sector unions, or lobbyists).
For what it's worth, here's how Claude summarizes the two cited papers.
Both Papers Agree on the Core Empirical Finding
Both papers conclude that the original Korpi-Palme paradox no longer holds as a robust empirical generalization.
Gugushvili (2021): "Recent studies have failed to confirm the existence of this paradox"
Marx et al. (2013): "This no longer holds as a robust empirical generalisation"
The Apparent Contradiction is About Interpretation, Not Facts
Gugushvili (2021) takes a more theoretically conservative stance:
Argues the theory was sound but measurement was flawed
Suggests we need better ways to study universalism vs. targeting
Maintains that universalism is still theoretically superior
Focuses on preserving the original theoretical framework
Marx et al. (2013) takes a more empirically driven stance:
Shows that when methodology is improved, targeting often performs better
Demonstrates this across multiple datasets and specifications
Argues the relationship has genuinely changed, not just measurement issues
Focuses on what the data actually shows now
They're Examining Different Time Periods and Scope
Gugushvili: Reviews 20+ years of studies testing the original theory, synthesizing contradictory findings
Marx et al.: Conducts original empirical analysis on mid-2000s data with expanded country coverage
The Resolution: Both Support a Nuanced View
The real conclusion from both papers combined is:
The simple universalism > targeting rule no longer works
Context and design matter more than universal vs. targeted labels
Modern targeted programs are different from the old-style means testing that Korpi-Palme criticized
"Targeting within universalism" may be optimal - generous systems that focus resources effectively
Bottom line: Both papers actually support the same practical conclusion - well-designed, adequately funded means testing can work effectively. The difference is that Gugushvili wants to preserve the theoretical framework while Marx et al. are more willing to abandon it based on the empirical evidence.
Neither paper supports simple, punitive means testing, but both suggest that modern, well-designed targeted programs can achieve strong redistributive outcomes.
Means tested programs are designed to steal from the middle class. They end up paying taxes and not getting benefits. Universal programs are better because the difference between what you pay and what you get tends to be lower.
Example
Means Tested for middle class:
Taxes = 80
Benefits = 0
Net = -80
Universal for middle class:
Taxes = 100
Benefits = 70
Net = -30
And I'm being generous there. Perhaps taxes are the same in those scenarios.
It's a libertarian fantasy that cutting universal benefits will lead to dramatically lower taxes. Governments maximize tax revenue, any reduced spending burden on universal benefits will get re-directed to welfare (for the poor, public sector unions, or lobbyists).
Why do you describe it as "from the middle class" rather than both the middle and upper class?
For what it's worth, here's how Claude summarizes the two cited papers.
Both Papers Agree on the Core Empirical Finding
Both papers conclude that the original Korpi-Palme paradox no longer holds as a robust empirical generalization.
Gugushvili (2021): "Recent studies have failed to confirm the existence of this paradox"
Marx et al. (2013): "This no longer holds as a robust empirical generalisation"
The Apparent Contradiction is About Interpretation, Not Facts
Gugushvili (2021) takes a more theoretically conservative stance:
Argues the theory was sound but measurement was flawed
Suggests we need better ways to study universalism vs. targeting
Maintains that universalism is still theoretically superior
Focuses on preserving the original theoretical framework
Marx et al. (2013) takes a more empirically driven stance:
Shows that when methodology is improved, targeting often performs better
Demonstrates this across multiple datasets and specifications
Argues the relationship has genuinely changed, not just measurement issues
Focuses on what the data actually shows now
They're Examining Different Time Periods and Scope
Gugushvili: Reviews 20+ years of studies testing the original theory, synthesizing contradictory findings
Marx et al.: Conducts original empirical analysis on mid-2000s data with expanded country coverage
The Resolution: Both Support a Nuanced View
The real conclusion from both papers combined is:
The simple universalism > targeting rule no longer works
Context and design matter more than universal vs. targeted labels
Modern targeted programs are different from the old-style means testing that Korpi-Palme criticized
"Targeting within universalism" may be optimal - generous systems that focus resources effectively
Bottom line: Both papers actually support the same practical conclusion - well-designed, adequately funded means testing can work effectively. The difference is that Gugushvili wants to preserve the theoretical framework while Marx et al. are more willing to abandon it based on the empirical evidence.
Neither paper supports simple, punitive means testing, but both suggest that modern, well-designed targeted programs can achieve strong redistributive outcomes.