Assuming the input impedance is the same. What happens in the case of an overload? CAT III 1000 V standard overload test is an 8 kV transient, so lets assume 16 kV for 2000 V (no idea if there's actually such a thing as CAT III 2000 V, let alone what the test conditions are). The MOV in one of the meters will start clamping the transient. This essentially shorts the inputs of that DMM, and puts almost the full transient across the second DMM, which might catastrophically fails since it suddenly gets to dissipate twice the voltage transient it was designed for.
This assumes the meters were properly designed, like the Fluke 27 that Dave recently reviewed. I wouldn't even trust some random cheap DMM which skimped on safety features, eg. Uni-T, to handle its rated transient, let alone twice that.