With Suzanne Berger, we mourn the decline in U.S. manufacturing and recognize the depth of its problems, revealed by both massive job losses and flatlined productivity growth since the late 1990s. We agree that American finance has both contributed to this decline and made fixing it more difficult. We also share Berger’s skepticism about growth in good manufacturing jobs absent changes in public policy. The celebrated wave of manufacturing onshoring is a feeble PR hoax; the “wave” is only a trickle of generally lousy jobs. Nor will natural gas abundance or rising Chinese wages save us. Only about a tenth of U.S. manufacturing (mostly plastics) is fuel-dependent enough for the former to make a difference, and Chinese wages can keep rising for a long time before approaching those here.

We disagree, however, with Berger’s emphasis on innovation as the primary justification for creating more good manufacturing jobs. Innovation is important, of course, and a domestic manufacturing commons would likely enhance it. But we have no idea just how large such a commons need be to support innovation, and we doubt it is the most efficient way to generate product breakthroughs. Berger herself notes that companies such as Apple and Cisco have strong records of domestic innovation despite offshore production. And the viability of this innovation-without-production model is not limited to Berger’s examples from electronics. For instance, as fuel-economy regulation drives automakers toward a common set of global platforms, small cars and trucks are increasingly designed in Europe and Japan, even as more are assembled here.

Our government chooses not to revive manufacturing.

The best justification for creating more good manufacturing jobs lies elsewhere: in employment for people without graduate degrees and the positive effect such jobs would have on the rest of the labor market. We think the decline of manufacturing has been driven less by the financial dissolution of vertically integrated manufacturers than by those firms’ offshoring of production or exit from competitive market segments. This fostered the subsequent rise in imports without compensating exports, which in turn drove down wages domestically and helped to destroy the middle class. We think good manufacturing jobs can help to fix that.

How might we make domestic manufacturing high-performing enough both to displace imports and pay well? The answer starts with understanding where the remaining good jobs are. Census data show that fewer than 4,000 large firms (defined as 500 or more employees) account for 55 percent of manufacturing jobs and 62 percent of payroll. On average, they pay $25 an hour. Firms with fewer than 500 employees average $19 hourly but often pay much less. That is why median pay for all U.S. hourly production workers was just $14.08 in 2012—better than retail and food service but no better than the rest of the economy.

What would it take to get the most productive and highest-paying quarter of these manufacturers to account for, say, two-thirds rather than one-third of domestic output? We are talking about influencing a small number of firms. Just fifty of them do half the exporting and control more than a third of the imports. How could we get them to produce more? Like Berger, we doubt that new manufacturing technology centers and other public-private partnerships can or will do much to answer this question. Indeed, they don’t really even ask it. Instead, we advocate policies that significantly raise demand for the output of good firms.

The only plausible source of such demand is government. Happily, there is immense public need for infrastructure and other public goods: water and sewer pipes and pumps, structural steel for roads and bridges, transit vehicles for cities, high-speed trains, a smarter and more reliable grid. This is not the vanguard of innovation, but it is stuff that can be made better, with built-in digital technology to monitor performance. The best way to build the industrial ecology Berger seeks would be to open such projects to bidding but require eligible firms to manufacture domestically, train and pay well (say, at least $25 per hour), meet the highest international standards on cost and quality, and share intellectual property with high-performing suppliers.

Such an effort would require political leadership in short supply these days. The mechanics of it are not so hard, though. In fact, this model is already in place. Large, vertically integrated, well-paying, and competitive manufacturing firms still exist, from Lockheed-Martin, Boeing, and Raytheon in the United States, to Daimler and Siemens in Germany. What all these firms have in common is that they are major government contractors. The difference between our situation and Germany’s is not that our government could not revive domestic manufacturing but that it chooses not to. We have chosen not to impose rigorous standards on our public contractors, and we ask them to manufacture tools for surveillance and war, not more pedestrian and pacific, but urgently needed, domestic public goods.