In a workshop on cost estimates at a conference on paid family leave last winter, one presenter said that in his model he chose assumptions that would result in a total cost figure at the high end of the possible range in costs for a new program. That way, if his state adopted the program, it would be funded at a high enough level to be certain of not running out of money.

The next presenter, in contrast, said that he chose assumptions that would put his cost estimates at the lower end of the possible range. That way, legislators would be far more likely to consider and ultimately pass a program for paid family leave. He reasoned that once the program was in place, its political popularity and far reaching benefits would assure continued funding at whatever level proved necessary.

Both of these presenters were serious about making realistic cost estimates. Both were careful to consider a broad range of issues that would affect need, take-up rates, average lengths of leave, administrative costs, etc., and to use the best data available in making their estimates. Both were also supporters of paid family leave, and thus not trying to sabotage enactment of the policies. Paid family leave has not been adopted in either of their states – or any state to date, so we do not have an ending to the story.

However, this example does illustrate what we all know to be true at some level – that cost and benefit analyses are not neutral, no matter how scientific our approach, and that our analyses often are used in very political ways.