Previous studies have shown that the most efficient economic system causes inequality to grow. Policies that reduce inequality are believed to have a trade-off: If you punish people too much for doing well—by reducing their incentives to pass on wealth to their children—these people will reduce their effort, to the detriment of the economy as a whole. But allowing inequality to increase concentrates the world’s wealth into fewer and fewer hands.Well, that jibes with everything I've always believed, so of course I'm going to find this interesting. Mankiw seems less convinced. Either way, here are the two papers where Werning lays out the case for a progressive estate tax (along with, in fact, subsidies for bequests by lower-income families). Here's a paper in which he rather cleverly rebuts the notion that the state needs to cut off unemployment benefits after six months in order to spur workers into finding jobs.
Werning found that the models at the core of these judgments were incomplete. Allowing inequality to grow, unfettered, is economically optimal only if one looks at just the first generation. But if you take into account the children of first-generation parents, and their children’s children, then the most preferable system is not one that allows inequality to grow, but one that attempts to stabilize the distribution of wealth.
His paper shows that the transmission of wealth should be regulated to prevent an accumulation of luck—that children should essentially be insured against the family into which they are born.