"And actually, over the last 10 years, tort filings are down."
I keep hearing this, (for example, in this comments section on the medical-malpractice debate) so I looked it up (via Childs). When we talk about liability reform, it's product liability that gets the crux of attention: that's where the billions of dollars get sapped out of the economy, where innovation gets squeezed and punished, where plaintiff fraud is most common.
Federal non-asbestos product liability suits, 1990: 5,812
Federal non-asbestos product liability suits, 1995: 21,310
Federal non-asbestos product liability suits, 2005: 29,052
That 1995 number is unusually high because that's the year that most of the breast implant cases were filed.
Even gerrymandering it to the last ten years, it's not true that tort filings are down, but if you change the baseline to 1990, it's not even close to true. The asbestos number has some impact: in 1990, there were 13,809 asbestos cases; in 1995, 7,187; in 2002, 26,818 new asbestos cases; but only 1,243 in 2005—a lump influx that distorts the debate in each direction, depending on how you choose your baseline. (I'm quite confident we'll see a lot of anti-reform studies in the next decade that use 2002 as a baseyear, and announce that litigation is going down.) It may just be that "one year" is not a good time period to use to measure litigation trends. It's not even clear that "cases" is the appropriate measure: the $80,000 product liability case of a single plaintiff is grouped in with the consolidated ten-plaintiff Vioxx case filed in the southern district of Illinois where each plaintiff will seek millions of dollars as a single case each.
Updated to add: Of course, not all product-liability suits are federal filings; most are originally filed in state court. While defendants will normally remove suits if there is a colorable argument to do so (thus adding to the federal statistics, even though many of these cases are remanded back to state court), a good number of cases are filed in state court and stay in state court without a removal being attempted. (E.g., cases filed in the defendant's state of domicile; cases with a reasonable same-state co-defendant such as auto-accident cases.) State court data doesn't break these out.
Even asbestos cases don't distinguish between cases with one defendant and cases with 106 defendants.