The Rhythm of EU Law-Making (Part 1)


This is the first post in a series where I explore the EU’s legislative process using data from the past five years. Don’t worry if statistics or EU procedure sound intimidating – I’ve written this in a way that requires little knowledge of either.


How fast – or slow – does the EU really move when adopting laws? Two extreme examples:

  • Exceptional macro-financial assistance to Ukraine: 12 days from the publication of the legislative proposal to the final law.
  • Directive on improving the gender balance among company directors: 3675 days from proposal to law.

Considering the difference of about 10 years, does EU law-making follow any discernible rhythm, or is each law dancing to its own tune? In this and the next posts, I examine how long it usually takes lawmakers to adopt a proposal, and whether there are certain groups of laws that move at similar speeds through the legislative process.

I use data on 548 legislative proposals that the European Parliament and the Council adopted, or at least started to consider, during the last mandate (2019-2024). These represent nearly all of the laws that moved through the Ordinary Legislative Procedure (OLP), the EU’s standard method to adopt laws, over the past five years.

Scanning for a heartbeat

At a very high level, a timeline of the EU’s legislative process could look like this. Basically, I focus on a few key events and count how many days elapsed between the publication of the legislative proposal and each step.

Using the data in this way allows me to plot all steps of every law into one figure. The ridgelines reveal where observations are clustered, i.e., where laws took a similar amount of time to reach a step in the legislative process. I exclude a few extreme values above 1500 days (e.g., for the law on gender balance) to make it legible.

What looks like barcodes under each ridge show when each law reached this step. The black line bisecting each ridge represents the median, meaning half of all observations are below and half are above this value.

Let’s consider three key characteristics of the data:

1. Not all proposals have been adopted

Not all 548 proposals have been adopted yet, which affects our analysis. If we focus only on the 390 adopted laws (i.e., we ignore proposals that are still moving through the process or got stuck on the way), the calculated median and mean decrease. This makes the process appear faster than it really is, especially at the early stages, because the median and mean shift to the left.

However, if we include all proposals in the analysis, the median and mean shift right – especially at the early stages where we add most observations. (The figure above includes all proposals.)

To demonstrate why the data selection matters for the results: Parliament’s calculations from March 2023 show that the average time to adopt an OLP file is 13 months (approximately 390 days). This is accurate and can be reproduced using my data, even though my results are significantly higher, with an average duration of 618 days and a median of 537 days. (A few extreme values inflate the average, remember the law taking 3675 days.)

The discrepancy between Parliament’s and my result arises from differences in the time period and the number of files considered. Parliament officials based their calculation on data from the first three and a half years of the last term and around 160 adopted laws. In contrast, I use data for the full term and 390 adopted laws. However, my values are close to Parliament’s calculation for the full 8th term (2014-2019), during which the average time to adopt an OLP law was 18 months (approximately 540 days).

2. Proposals follow different paths

The median values generally increase from one step to the next, which is expected. But how do we explain cases where the median decreases?

The answer lies in the fact that not all laws move through every stage, leading to a varying number of observations at each step. This is not only because some proposals are still waiting to be adopted. More importantly, even though all proposals follow the Ordinary Legislative Procedure, their exact path still differs. For instance, some laws did not require lengthy negotiations between the Parliament and the Council (trilogues) because both institutions accepted the Commission’s proposal without amendments.

As a result, there are several groups of proposals that follow slightly different paths. If you look closely, you will notice early peaks at stages 3, 7, and 8. These peaks represent a cluster of urgent laws, many of which were adopted within the first 100 days.

We need to account for these different paths because they significantly impact the (median and average) duration of each stage. I’ll explain these paths in more detail in the following posts.

3. More variation at the later stages

If you have recently been to the beach, you might spot the third characteristic of the data: the values are more dispersed at the later stages. This creates a visual effect similar to a wave rolling onto the shore – as we move from step to step, peaks flatten and the ridge expands to the right.

This increasing dispersion is expected. The time that proposals spend at each stage depends on a variety of factors, such as political support, legal complexity, whether they arrive shortly before the summer break, etc. I do not control for these here.

These three data characteristics – variation in current status, different paths to adoption, and significant dispersion – should caution us against placing too much emphasis on median or average values calculated across all 548 files. While these metrics are useful for context, they ignore a lot of complexity. I include them here for information, but there is a clear need to delve deeper to understand which files move through the process at what speed.

Procedure stepMedian (in days)Mean (in days)**
1. Committee(s) assigned4692
2. Committee report/decision without report254  287  
3. Decision by Parliament before trilogues (1st reading)245  249
4. Decision to enter trilogues337393
5. Committee confirms political agreement462569
6. Decision by Parliament after trilogues (1st reading)513612
7. Council adoption (1st reading)*408  500
8. Act published537618
Note: Calculated across all proposals.
(*) Parliament’s data on Council adoption has some gaps.
(**) A few extreme values inflate averages. The figure does not show values above 1500 days, but they are included in the calculations.

Plan for the next posts

I hope this overview of the EU’s legislative activity over the past five years has sparked your interest in more data-driven analysis. To me, the numbers suggest that there are some underlying patterns that are worth exploring further.

In the next post, I will present the first detailed timeline for a group of urgent and relatively uncontroversial laws (such as assistance to Ukraine and Covid-related measures) that were adopted quickly.

In the third post, I focus on the now dominant path to adopt OLP files, which involves detailed negotiations (trilogues) between the EU institutions.


Be the first to know about the next post:


5 responses to “The Rhythm of EU Law-Making (Part 1)”

  1. […] the EU’s legislative process using data from the past five years. I recommend you start with part 1 and part […]

    Like

  2. […] Part 1 introducing the data and visualisation […]

    Like

  3. […] To help you understand the sequence of events, we use a relatively uniform timeline across the figure. While the precise timings reflect the peculiarities of the CLP revision, the figure offers you a better sense of the general pace of EU law-making. (I explore the rhythm of the legislative process in more depth here.) […]

    Like

  4. […] the EU’s legislative process and build a timeline of its key steps based on a large dataset. (Part 1, Part 2, Part 3, summary & […]

    Like