Heterogeneous Effects. Manufacturing firms
Heterogeneous Effects. Data The database used to analyse possible heterogeneous effects with respect to different network or traffic characteristics contains information on each individual train circulated on the railway network in the period 2018-2022. For each train it is known the origin and destination, date and time of departure, railway undertaking, average weight, category, TAC (with the specification of both component A and component B), kilometres travelled, commercial speed, and kilometres travelled by dangerous goods. Furthermore, other information on route and/or network characteristics are included to this database. In particular, two dichotomous variables are included to indicate whether a train is combined or conventional and whether a train travels mostly during the day or at night. Furthermore, it is considered whether the train travels on a high-performing line or not, whether the origin and/or destination coincides with a port or an intermodal node, and the day in which the trains start and end their trips (weekday/festive). It is worth noting that the database covers a fairly limited period of time and the only change in TAC recorded over the period is that related to the lowering or cancellation of part of the TAC as a result of the temporary measures applied due to the COVID-19 pandemic. In addition, in the years analysed, there is no phenomenon of the opposite sign, i.e. a large increase in TAC or in its components. The database is reorganised so that the total train-kilometres circulated on each route can be observed in each year.7 Accordingly, after calculating the trains-kilometres of each individual train, these are aggregated at route level, and also the kilometres travelled by dangerous goods and the TAC paid (summing-up both component A and component B) are aggregated at the same level. The kilometre length of each route, on the other hand, is calculated as the average of the kilometres travelled by each train on a specific route.8 Also other variables related to individual trains are aggregated at route level by computing the average value.9 Finally, in order to conduct the analysis, the overall TAC (and its component A and component B) per train and per tonne is computed. Table 5 shows the relevant descriptive statistics. 7 Note that in this case the route is sensitive to the difference between origin and destination, so that a route from A to B differs from a route from B to A.
Heterogeneous Effects
