PostEurop welcomes the opportunity to send input after the 2nd workshop on the Prospective study on the future of the postal sector held on 27 March 2024.

Process

PostEurop welcomed the opportunity to further discuss the baseline scenario 1 during the workshop of March 27th. It is however unclear to us, whether we will get the opportunity to also further discuss and comment on the other 4 scenarios before they are presented in their final version?

Data collection

With regard to the collection of data and information, when making use for your desk research of the latest CE Market Monitoring report 2017-2021, it is important for you to note that PostEurop has raised specific concerns with regard to chapter 7 of the study, concerning the cross-border parcel delivery market. Please find here the letter we sent to Copenhagen Economics and the European Commission with our argumentation on why the conclusions of this chapter should be used with caution.

Methodology

The aim of the study is extremely important for the postal sector because it will support the European Commission in the development of a prospective assessment of the postal sector, including the potential need for and impacts of revisions to the EU postal services legal framework (slide 3). Thus, this study will shape the sector for the next two decades at least.

Hence, special attention should be placed on the methodology used, and as a subset on 1- the data used in the model and 2- the predictions for the future; 3- the model used and 4- the assumptions behind the modelling.

We will start with the data used and the projections for the future. As explained in slide 6, the quantification of the indicators will consist of two steps: 1- historical data for each indicator are analysed and extrapolated to quantify the future values of current trends, 2- understand where indicators may deviate from the extrapolated trend and to propose appropriate adjustments.

As discussed in slide 11, the data will be obtained from a number of sources, but the focus of the study will be on four main sources: UPU Database, DG GROW Postal services database, Copenhagen Economics (2022) and Eurostat. Some of these databases are not updated (Database from DG GROW goes up from 2012-2021, Copenhagen Economics 2021 (slides 13-14)). Revenues, profits, employment refer to 2021, maximum 2022 (slide 15). Even more important is the fact that, as mentioned in slide 20, they are based on historical data that are heavily impacted by Covid-19 thus the extrapolations of these data are heavily impacted by the events taking place in 2020 with a tail in 2021.

The pragmatic approach of a linear extrapolation by default (in some cases exponential and log extrapolation) risks overestimating events such as the growth  of  e-commerce (as discussed in slide 23-24).

The simple linear extrapolation of recent historical data on employment (slide 25) also is not justifiable if the growth of parcels will not be as dramatic as in slide 24, thus there is a risk of no coherence between the same scenario. Furthermore, the extrapolation of historical prices is even more absurd because most of these prices are regulated (slide 27 and 28) and the extrapolation of the number of enterprises providing postal services as a proxy of the degree of competition does not take into account the peculiar markets and the importance of the national regulatory architecture defining the rules to obtain licenses (regional, national) to operate in a market, risking the comparison of apples and oranges. Overall, the adjustment of past trends (slides 31 to 33) are minor adjustments that do not influence the trends.

Concerning the model used and the assumption of the variables of the draft assessment model, we appreciate the use of econometric approach to determine (non-casual) relationships (slide 36 and 40). However, we question several issues.

First of all, the NERA (2004) model is a simplified cost minimization model which means that the focus is to reduce costs and the aim is cost efficiency (a strong assumption because it states that the model aims at one thing; reducing the USO burden) but we understand that this is the approach which has been used by the literature in our industry. However, there is a lot of literature on more complicated models applied to our sector presented in the Toulouse School of Economics and on the Rutgers/EUI Conferences in the past two decades. We think that the objective should be to provide a robust method for assessing the sector which will be useful for the next two decades. This study should build on available literature on more complicated models, earlier work like the Principal Component Analysis to identify the determinants of demand, then utilize parametric, non-parametric (DEA) and semi-parametric methods to analyse efficiency models.

Secondly, a simplified model does not take into account many indicators relevant especially for medium and small operators. For example, mail per household is more important sometimes than total mail and it aims to find a correlation between % urban population with total costs and not unitary cost. European operators have different dimensions in terms of volumes, costs, population that should be taken into consideration.

Furthermore, studies from other sectors like Railway (Farsi, Filippini and Greene, 2004, Efficiency Measurement in Network industries. Applied to the Swiss Railway Companies) have taken into account also some indicators of the orography of the territory where a network industry operates. For example, there are countries with complicated geographical situations (mountainous areas, islands, scarcely populated areas, overseas territories) which impact the cost structure of postal operators (without complicating the equations there are solutions to incorporate these indicators).

Finally, what might need some clarification in the methodology is how to include in the extrapolation for future trends and in the econometric models, the changes in the regulatory architecture. This topic raises an issue: how will the impact of the regulatory changes on the dependent and independent variables be determined (which are the assumptions behind the proposed changes)?