A Range of Approaches to Procurement for COVID-19

Shortages of medical products, including but not limited to ventilators, personal protective equipment, and nasal swabs, continue to be a danger. If the number of COVID-19 cases continues to increase exponentially for weeks before peaking, many lives may be lost as a result. In a series of posts, I have applied an approach described in a paper in progress to the specific problem of ramping up ventilator production. But for this post, let me place the specifics of that proposal aside to canvas the range of options for increasing medical production.

The Pure Market Approach. Alex Tabarrok argues that we should let the markets work. As he points out, ventilators and other scarce medical equipment are extremely valuable right now, and that value drives up prices, and the price signal creates incentives for production. In response to concerns about price gouging, he states,"It's very important that manufacturer prices be allowed to rise to reflect true scarcities and to get resources flowing in the right direction." I agree, but Tabarrok concludes, "So far, we are doing that and the system is working well." Are we? I believe that if manufacturers created needed medical products and charged five times the usual rate, they would be vilified as price gougers and maybe even prosecuted. But that multiplier or maybe even a higher one might be necessary to provide incentives for creation of new production lines that produce machines that might be useful for only a limited time, if caseloads continue to increase and no alternative treatment emerges. The market will work in the sense that production will go up somewhat, but under the constraint that prices can't rise much. Production thus will not ramp up nearly as much as it would in a genuinely free market. The fact that we haven't seen alleged price gouging on new ventilator contracts suggests to me that the market is not working. The pure market approach seems likely to work only if the government were to pass new laws clearly indicating that producers can charge whatever the market will bear, without vilification. That seems unlikely the day after a new executive order preventing price gouging and hoarding of medical supplies.

State Procurement. President Trump has argued that the federal government isn't a "shipping clerk," and thus that states should procure what they need. State procurement is not without virtues in this context. States can be "laboratories of experimentation" in taking different approaches, both with respect to what they wish to buy and with respect to how they go about approaching the purchasing. Moreover, states competing with one another might bid up the prices of needed resources. If one's goal is to stimulate production, this is good; perhaps there will be fewer complaints about price-gouging if states (rather than individuals) are making high offers, and, as noted above, high prices may be necessary to stimulate extraordinary production. But there are downsides. States have limited expertise and resources to think comprehensively about how to design a procurement system. Different states may demand different standards, preventing production at scale. States may learn from one another only very slowly, and speed is of the essence. Some have criticized state procurement because it will not lead to production going to where it is most needed (although rigorous application of that criterion suggests that perhaps we should ramp up production and ship off everything we produce to other countries). Some states may end up with excess capacity and other states with insufficient capacity, and states may be hesitant to sell their excess capacity to others, both because of the chance that they may lead it later and because criticism will prevent them from selling at market prices.

The Defense Production Act. The government can simply order private manufacture of particular goods, although "there is a genuine question as to whether the law can be used to compel a company to accept a new contract for a product or service it does not ordinarily provide." Even assuming legality, however, the government may not know which companies are best positioned to produce which goods. Moreover, forcing a company to produce may not lead to the company having incentives to produce as efficiently as possible. When the company faces a bottleneck, for example because of a shortage of an input, it may just do nothing, rather than what it would do in a market system, namely seek out alternative suppliers, bid up the price of the input, or produce the input itself. The government can respond by forcing some other supplier to produce the input. This requires the government to perform well in its central planning, and even with the best planning, reluctant producers may not perform well. Carrots are generally better than sticks.

Moral Suasion. President Trump has encouraged automobile manufacturers to produce. This has the same problems as the Defense Production Act. Do we think the government can choose the right producers, identify what is needed, and motivate them properly? But when the only stick is the possibility of bad publicity, the approach may be even less effective. Elon Musk initially promised to convert factories "if needed" (by which time it might be too late), but then bought himself some good publicity by importing some ventilators from China. That is certainly a useful contribution from the United States's perspective, but the key is to increase manufacturing capacity, either domestically or internationally if other countries will permit such goods to be exported.

Federal Procurement. My colleague Chris Yukins offers a powerful argument for federal procurement generally, while calling the Defense Production Act a "risky prospect." The federal government, Yukins argues, has "nearly unlimited resources–its commitment to pay, and the world's largest procurement apparatus–to drive procurement of equipment that will save lives." Central to his argument is that federal contracting always includes a clause allowing the government to "terminate for convenience." It might seem that this would be exactly what we would not want, because producers may be concerned that they will be left with nothing if the pandemic abates. But Yukins points out that when a contract is terminated for convenience, companies' sunk costs will be covered by the United States. Thus, one might think, contractors have nothing to lose—at worst, their costs will repaid, and at best, they will receive the contracted-for price. Unfortunately, however, these sunk costs are not likely to include lost opportunity costs. Conversion of a GM plant from cars to ventilators not only costs money, but also means that GM will manufacture fewer cars. There are also additional reasons to worry. First, will the government be able to conduct a procurement sufficiently quickly? Second, does the government have enough information to choose wisely and then make adjustments as necessary? There are many plausible designs for ventilators, including for example designs in which multiple patients share a single ventilator. The optimal design may depend on the number of cases, with simpler and cheaper designs to be preferred, the worse things get.

Cost-Plus Pricing. Ezekiel Emanuel also suggests federal procurement, specifically endorsing the use of cost-plus pricing. He cites an article by Utpal Dholakia, who argues that one virtue of this approach is that it is simple to implement and to explain. But Dholakia also acknowledges that cost-plus pricing "discourages efficiency and cost containment," especially for stand-alone projects. Suppose that the government simply promised to pay costs plus 10% for all new ventilator production in the next six months. A producer then has no incentive to be restrained on input price; indeed, the more that is paid for inputs, the better, at least up until the point where the government decides that the price was inflated, a difficult assessment to make in a time when many inputs face shortages. Meanwhile, there is little incentive to make sensible decisions about just what ventilator models to produce, and opportunity costs are not covered, so many of the objections above still apply.

Cost-Plus-Performance Pricing. In a recent article, I discussed various recent proposals to use cost-plus measures for calculating damages in patent cases. Some of the problems that I identified apply here as well, such as the danger that incentives will be too low if incentives are set too low and the danger that too much will be spent. But I offered a simple model that suggested that it might make sense for patent damages to be based in part on a cost-plus calculation and in part based on traditional considerations. The intuition is that if only half of what one receives is based on a cost-plus calculation, then the incentive to overspend is eliminated or at least greatly diminished. At the same time, if patent damages sometimes are either windfalls or insufficiently compensatory, having damages partly determined by cost-plus considerations should diminish this concern.  Much the same could be true for production of ventilators and other supplies. If we promise to compensate producers for some of their costs but make their overall payment depend on performance (for example, measured in QALYs saved), then we can improve predictability while still giving incentives for performance.

There are many ways something like this could work, but let me suggest just one. Like my previous proposal, this involves ex post assessment of performance, but I will place aside the element of random selection. Suppose that the government commits $100 billion to production of needed medical equipment, and it wants ten companies to be competing to produce equipment. Those companies could be selected at auction, with the auction proceeds added to the fund to be distributed to the companies. Companies can be required to release information about their spending commitments as those commitments accrue, so that all of the competitors can have a good sense of what is happening in the market and how much it makes sense to invest in different approaches. When all is said and done, the government will measure the performance of each of the ten companies in terms of QALYs saved, and it will calculate cost figures for each company. It will then distribute $50 billion based on the performance measure and $50 billion based on the cost figures.

A system like this requires minimal government oversight ex ante and thus could be created quickly. The auction mechanism helps ensure that the companies best situated to perform will be chosen and that those chosen have a genuine commitment to an aggressive effort. The compensation for performance gives companies incentives to save lives and to identify the types of goods and approaches that are most needed. Finally, the compensation for costs reduces the risk to each company. My own view is that this last component is not essential and that the entire contracting process could be based on performance, because capital markets can effectively absorb risk, but this is a concession to those who believe that the contracting process demands greater predictability. Of course, one could adjust the portion of the fund to be allocated to cost and the portion to be allocated to performance.

The bottom line is that we need production. Any of the approaches discussed here is better than nothing. And approaches that can generate production quickly and that can harness incentives for private sector creativity should be preferred. I fear that it is too late to make a meaningful impact on cases over the crucial next few months. But it is not too early to identify better approaches for the next emergency.