Piano Forum

La campanella would likely have been played at a similar tempo to the original piece it's based on (the last movement Paganini's Violin Concerto No. 2) so as to be recogniseable. Although Paganini did not supply metronome marks either, the concerto rondo is actually taken at a quite slow tempo—"Andantino allegretto moderato" per the first edition—so where a standard Allegretto 6/8 might be taken at dotted quarter = 80 or faster, Paganini probably intended a tempo in the range of 60 to 72. A similar tempo range would work for the main section of the etude as well, and allow the 32nd notes to speak clearly (of course the tempo must then be accelerated significantly, as indicated).

For most other pieces by Liszt not based on pre-existing material, you have to make your own judgments (he did assign metronome marks to some of his late works). e.g. the Presto at the end of Hungarian Rhapsody 6 should almost certainly be played as fast as is comfortable, but that speed will vary depending on whether you are György Cziffra or an ordinary human being. The only potentially useful information I've gathered about Liszt and tempo is that he initially trained by practicing to a metronome (and his principal teacher, Czerny, was extremely insistent about the need to find the tempo giusto for a work that brings out its character) and therefore almost always played in strict time, with any changes in tempo being so smooth as to be imperceptible. So essentially, once you've determined what tempo works best for a particular passage, Liszt would have expected you to stick to it and make sure any rubato or tempo changes sound natural and unnoticeable.

The Neue-Liszt-Ausgabe does offer tempo suggestions from Liszt's students in a few works where he did not supply them himself (e.g. the Bénédiction de Dieu dans la solitude, Funérailles, etc)