Who Really Bears Labour Taxes? – Dr. Dudi Duta Akbar

A Meta-Regression Methodology with Explicit Economic Models

Main reference
Melguizo, Á., & González-Páramo, J. M. (2013). Who bears labour taxes and social contributions? A meta-analysis approach. SERIEs, 4, 247–271. Who bears labour taxes and soci…

1. Methodological Core: From Wage Equations to Meta-Regression

This study does not estimate the incidence of labour taxes using a single country or dataset.
Instead, it applies meta-regression analysis (MRA) to synthesize dozens of empirical estimates reported across the international literature.

The methodological objective is twofold:

To estimate a consensus parameter measuring how labour taxes affect net wages, and
To explain why empirical studies report widely divergent results, despite analysing the same economic mechanism.

2. Benchmark Empirical Model in the Literature

The studies surveyed in the meta-analysis are largely based on a common wage-setting framework, which can be written as: $f(W_{it}) = g\Big(P_{it},\ \frac{Y_{it}}{N_{it}},\ U_{it},\ h(TAX_{it}),\ X_{it}\Big) \tag{1}$

Variable definitions

$W_{it}$ : net nominal wage or labour cost (typically in logs)
$P_{it}$ : price level (output or consumption deflator)
$Y_{it}/N_{it}$ : labour productivity
$U_{it}$ : unemployment rate
$TAX_{it}$ : labour tax wedge (log-transformed)
$X_{it}$ : institutional and structural controls (unions, bargaining structure, EPL, benefits, etc.)

Functions $f(\cdot)$ and $h(\cdot)$ may be specified in levels or growth rates, depending on whether short-run or long-run effects are estimated.

3. Key Parameter: Wage Elasticity to Taxation

Across studies, the central object of interest is the elasticity of net wages with respect to labour taxation: $\beta = \frac{\partial \ln W}{\partial \ln TAX}$

This elasticity directly measures backward tax shifting:

$\beta = -1$ : full shifting (workers bear 100% of the tax burden)
$\beta = 0$ : no shifting (workers bear none of the burden)
$-1 < \beta < 0$ : partial shifting

This elasticity is therefore a natural and comparable effect size across empirical studies.

4. Harmonising Wage and Labour-Cost Specifications

Some studies specify the dependent variable as labour cost rather than net wages: $LC = w(1 + SSC)$

where $SSC$ denotes employer social security contributions.

Using this identity, the authors standardise all reported estimates into equivalent net-wage elasticities, ensuring cross-study comparability. Who bears labour taxes and soci…

5. Formal Definition of the Tax Wedge

A crucial source of heterogeneity lies in how taxation is defined. The study explicitly models alternative tax wedge constructions.

Let:

$p_{it}$ : effective personal income tax rate
$essc$ : employee social security contribution rate
$fssc$ : employer social security contribution rate
$c_t$ : consumption tax rate
$pp$ : domestic price deflator
$pp^*$ : international price deflator
$a$ : share of domestic goods in consumption
$e$ : exchange rate

(a) Salary wedge (taxes + price wedge)

$\text{Salary Wedge} = \frac{(1+c_t)(1+fssc)}{(1-p_{it})(1-essc)} \cdot \left(\frac{e\,pp^*}{pp}\right)^{1-a}$

(b) Fiscal wedge (tax components only)

$\text{Fiscal Wedge} = \frac{(1+c_t)(1+fssc)}{(1-p_{it})(1-essc)}$

(c) Direct tax wedge

$\text{Direct Tax Wedge} = \frac{(1+fssc)}{(1-p_{it})(1-essc)}$

(d) Employer social contribution wedge

$\text{Employer SSC} = (1+fssc)$

These alternative definitions are explicitly introduced as moderators in the meta-regression.

6. Meta-Regression Model

Each empirical estimate extracted from the literature is treated as one observation in the meta-dataset. The meta-regression model is specified as: $\beta_s = b + \alpha_{s,i} + \alpha_{s,t} + \gamma X_s + u_s \tag{2}$

Interpretation

$\beta_s$ : reported wage elasticity to taxation in study $s$ s
$b$ : true elasticity (consensus estimate)
$\alpha_{s,i}$ : geographical fixed effects
$\alpha_{s,t}$ temporal fixed effects
$X_s$ : vector of moderators (institutions, tax definition, data type, estimation method)
$u_s$ : disturbance term

The coefficient $b$ b represents the average economic incidence of labour taxes, after controlling for methodological differences.

7. Key Moderators in the Meta-Analysis

(a) Time horizon: short-run vs long-run

Nominal rigidities imply gradual wage adjustment. This is captured via: $\beta_s = b + \delta \cdot \text{ShortRun}_s + \ldots$

A positive $\delta$ (given $b<0$ ) implies less shifting in the short run.

(b) Institutional regimes

The literature distinguishes between:

Anglo-Saxon economies
Continental/Mediterranean economies
Nordic economies

Formally: $\beta_s = b + \lambda \cdot \text{Nordic}_s + \ldots$

The Nordic regime is associated with stronger wage adjustment, reflecting coordinated bargaining and effective public sectors.

(c) Tax wedge composition

Including indirect taxes and price wedges systematically increases the estimated elasticity: $\beta_s = b + \theta \cdot \text{Salary/FiscalWedge}_s + \ldots$

8. Estimation Strategy

Given the presence of outliers and clustered estimates, the authors rely on:

Robust regression techniques, and
Weighted Least Squares (WLS), using citation-based quality weights.

This ensures that results are not driven by extreme observations or low-quality studies. Who bears labour taxes and soci…

9. Reading the Results in Structural Terms

The estimated consensus elasticity is approximately: $b \approx -0.70$

This implies that:

a 1% increase in labour taxation reduces net wages by about 0.7%,
meaning that workers bear roughly 70% of the tax burden in the long run.

In Nordic economies, this figure approaches full shifting, while short-run effects are substantially smaller.

10. Methodological Takeaway

This meta-analytic framework shows that tax incidence is not a fixed parameter, but a structural outcome shaped by:

wage-setting institutions,
tax design, and
adjustment dynamics.

By formalising the literature through explicit equations, the study moves beyond narrative surveys and delivers a quantitative consensus grounded in economic theory.

This methodology demonstrates how meta-regression can transform fragmented empirical evidence into a coherent structural interpretation of labour tax incidence—without relying on a single country, model, or dataset.