We employ a Structural Equation Modelling (SEM) approach to estimate how EU accession affects GDP growth and other development outcomes through various transmission channels. SEM is particularly well suited for this analysis, as it allows us to disentangle the pathways through which EU integration exerts its effects.

Concretely, we focus on three main channels:

• higher EU budget transfers,
• increased exports to the EU,
• institutional improvements.

The analysis is conducted in a panel setting covering the period from the late 1990s to 2023. It includes Bulgaria, Romania, and Croatia as benchmarks, alongside the six Western Balkan economies.

The concrete specifications for each of the eight simulated indicators are as follows:

• GDP per capita growthdepends on initial GDP (as in standard convergence models), EU budget transfers, exports to the EU, and control of corruption. These variables, in turn, are affected by EU accession.
• The change in income share of the bottom 20% is linked to EU budget transfers, which is linked to EU accession.
• The change in life expectancy is modelled as a function of GDP growth, which is influenced by EU accession through the channels explained above.
• The change in tertiary enrolment also depends on GDP growth, which itself depends on EU accession, as explained above.
• The change in energy intensity is modelled as a function of GDP growth, which in turn is tied to EU accession.
• The level of control of corruption is modelled to be affected directly by EU accession, as in one of the channels through which EU accession affects GDP growth.
• ICT exports are linked to GDP growth, EU transfers, control of corruption, and exports to the EU, all influenced by EU accession.
• Road density depends on EU budget transfers, which themselves depend on EU membership.

These specifications were determined empirically, by running different regressions with different combinations of variables, checking the significance of the coefficients.

The regression estimates are then used to generate simulations of the four scenarios. This is done by augmenting the rate of change from the baseline (status quo) scenario – which is the average over the last five years – with the estimated effect of the EU accession scenario. The effects differ for different scenarios, different indicators, and sometimes even for different countries (only in the GDP growth case, because the regression is in terms of the GDP growth, while the impact is in terms of GDP growth relative to the EU).

For example, for the GDP per capita growth indicator, the status quo scenario for Albania is that the country closes 1.9 pp of the gap to the EU every year. In the full EU membership scenario, this improves by 0.9 pp, to 2.8 pp per year. In the EU budget scenario, this improves by 0.5 pp, in the EU single market scenario by 0.26, and in the institutional reform scenario by 0.13 pp.

These augmented rates of change then translate into faster or slower years to EU. Specifically, if in the baseline scenario Albania is projected to reach the EU average level of GDP per capita PPS in 30 years, in the full EU membership scenario this drops to 20 years. in the access to EU budget scenarios, this become 23 years, in the access to EU single market, this becomes 26 years, and in the institutional reforms scenario, this is 28 years.