We address the task of explaining the effects of perturbations in procedural\ntext, an important test of process comprehension. Consider a passage describing\na rabbit's life-cycle: humans can easily explain the effect on the rabbit\npopulation if a female rabbit becomes ill -- i.e., the female rabbit would not\nbecome pregnant, and as a result not have babies leading to a decrease in\nrabbit population. We present QUARTET, a system that constructs such\nexplanations from paragraphs, by modeling the explanation task as a multitask\nlearning problem. QUARTET provides better explanations (based on the sentences\nin the procedural text) compared to several strong baselines on a recent\nprocess comprehension benchmark. We also present a surprising secondary effect:\nour model also achieves a new SOTA with a 7% absolute F1 improvement on a\ndownstream QA task. This illustrates that good explanations do not have to come\nat the expense of end task performance.\n