Table 2 shows the results of estimating the legacy of PNF membership in the 1920s across a range of categories of conflictual political events that took place during the Years of Lead in Italy. We report estimates for the variables included into the count model (the second equation) and not the inflating stage (probability of having a conflictual event in a province‐month).¹¹

All observations are defined at the province‐month level. In Column 1, the dependent variable considers all conflictual political events present in our dataset between 1969 and 1988, independently of the identity of the perpetrator. Column 2 includes all violent events in the dataset, also independently of the identity of perpetrator. All other columns refer to conflictual events—non‐violent (Column 4) and violent (Column 3)—perpetrated by neofascist actors.

We observe a positive and statistically significant effect of above‐average PNF membership in the 1920s on both conflictual and violent events that took place in Italy during the Years of Lead (Columns 1 and 2, respectively). The coefficient in the first column indicates that in provinces with strong PNF membership in the 1920s, the number of conflictual political events during the Years of Lead was 72.7 per cent 12 higher.

The estimated effect of PNF membership increases in magnitude and statistical significance for events perpetrated by neofascist organizations, and of violent nature. For instance, Column 3 shows an increase in the expected number of violent neofascist events in 1969–88 by 120 per cent in provinces with higher PNF membership in the 1920s. By contrast, the coefficient of the PNF legacy is not significant for non‐violent events (Column 4). Hence, our theory predicts violent legacies but not non‐violent legacies.

[…]

According to another alternative interpretation, neofascist violence during the Years of Lead might be a response to the violence that took place during the period preceding the formation of the fascist régime in Italy. Between 1920 and 1921 repeated violent clashes occurred between socialist and fascist groups. Memories of this period could have affected future cycles of violence.

To account for this, in Column 2 of Appendix Table A12, we add to the count model a variable capturing the total number of deaths in the political clashes that occurred between 1920 and 1921, as they might be systematically correlated with the early presence of PNF members. Our results are not substantially affected by the inclusion of this variable.

Further, we could observe neofascist violence during the Years of Lead as a response either to actions carried out by the government against far‐right political groups or to state repression against such groups in general. In Appendix Table A13, we analyse whether retaliation could be an alternative factor explaining the rise of neofascist violence in the 1970s and 1980s.

In Column 1, we reproduce the specification of Column 2 in Appendix Table A12 by adding to the first equation the total number of events of state repression against far‐right actors (at the province–month level).²²

The effect of higher PNF membership remains strong and highly significant. In the remaining columns of Appendix Table A13, we explore the legacy effect when the sample is split according to the cumulated level of state repression over the whole time period, to capture those areas where the activity of the state is systematically higher.²³

We observe a non‐monotonic trend in the effect of PNF membership: the impact of the fascist legacy seems to increase for low levels of repression, then drops when repression seems to reach its optimum level and it is effectively able to tackle the neofascist organizational structures that coordinate violence. After this point, we observe a strong backlash: excessive repression is likely to exacerbate tensions and magnifies the effect of the legacy, leading to a substantial increase in the level of far‐right violence.

Finally, local neofascist violence could also be explained with a logic of revenge resulting from the killings of fascist militants by partisans following Liberation Day in 1945, instead of early PNF membership. Liberation of Italy was completed on 25 April 1945, but in the following months left‐wing members of partisan bands caused about 10,000 deaths through extra‐judicial executions of individuals who were somehow considered linked to the fascist régime (Dondi 1999; Grandi 2013).

Given the brutality of these actions, neofascist violence during the Years of Lead might capture a long‐lasting retaliation for these killings, which are likely to coincide with early PNF membership. Such retaliation could also represent a channel through which the fascist legacy exacerbates the level of neofascist violence.

To evaluate this possibility, in Appendix Table A14 we add to the specifications of Table 2 and Appendix Tables A8 and A12 the number of extrajudicial deaths in a given province immediately after Liberation Day, together with its interaction with the dummy variable capturing strong fascist legacy.

The effect of the legacy stays strong and positive, whereas the independent effect of the killings on the level of neofascist violence appears to be negative, contradicting a potential retaliation argument. Retaliation does not seem to be a mediating factor either, as the interaction term is never statistically significant.