双语阅读｜机器智能定价“坑”消费者
MARTHA’S VINEYARD, an island off the coast of Massachusetts, is a favourite summer retreat for well-to-do Americans. A few years ago, visitors noticed that petrol prices were considerably higher than in nearby Cape Cod. Even those with deep pockets hate to be ripped off. A price-fixing suit was brought against four of the island’s petrol stations. The judges found no evidence of a conspiracy to raise prices, but they did note that the market was conducive to “tacit collusion” between retailers. In such circumstances, rival firms tend to come to an implicit understanding that boosts profits at the expense of consumers.
马萨诸塞州海岸附近的一个小岛玛莎葡萄园是美国有钱人喜爱的夏季度假胜地。几年前，游客们注意到这里的汽油价格大大地高于附近的科德角。即使那些有钱的人也讨厌这种欺诈行为。于是对该岛的四个加油站发起了一项要求统一价格的诉讼。法官们没有发现任何证据证明有共同涨价的行为，但他们确实注意到市场有利于汽油零售商之间的“默契合谋”。在这种情况下，竞争对手往往会达成默兑契，从消费者身上攫取利润。
No one went to jail. Whereas explicit collusion over prices is illegal, tacit collusion is not—though trustbusters attempt to forestall it by, for instance, blocking mergers that leave markets at the mercy of a handful of suppliers. But what if the conditions that foster such tacit collusion were to become widespread? A recent book* by Ariel Ezrachi and Maurice Stucke, two experts on competition policy, argues this is all too likely. As more and more purchases are made online, sellers rely increasingly on sophisticated algorithms to set prices. And algorithmic pricing, they argue, is a recipe for tacit collusion of the kind found on Martha’s Vineyard.
没有人因此而进监狱。 虽然暗地里串通定价是非法的，但是有默契地涨价却不是非法的——尽管反垄断人士试图阻止这种行为，例如，通过阻止合并行为，让市场容纳一定数量的供应商。 但是，如果促成这种默契定价的条件普遍存在呢？ 两位竞争政策专家阿里尔·埃兹拉奇（Ariel Ezrachi）和莫里斯·斯塔克（Maurice Stucke）在最近的一本书中认为，这是很有可能的。随着越来越多购买行为是在线完成，卖家越来越依靠精细的算法来定价。他们认为，算法定价会是玛莎葡萄园岛（Martha's Vineyard）上默契定价的翻版。
Consider the conditions that allow for tacit collusion. First, the market is concentrated and hard for others to enter. The petrol stations on the Vineyard were cut off from the mainland. Second, prices are transparent in a way that renders any attempt to steal business by lowering prices self-defeating. A price cut posted outside one petrol station will soon be matched by the others. And if one station raises prices, it can always cut them again if the others do not follow. Third, the product is a small-ticket and frequent purchase, such as petrol. Markets for such items are especially prone to tacit collusion, because the potential profits from “cheating” on an unspoken deal, before others can respond, are small.
允许默契定价的条件：首先，市场集中，难以进入。葡萄园的加油站远离大陆。 其次，价格是公开的，这让任何通过降低价格的方式来抢夺业务的想法都是自欺欺人。一个加油站降价很快就会被其他的加油站所效仿。如果一个加油站涨价，其他加油站不跟进，它只能再次降价。第三，该产品是高频小额，如汽油。这些产品的市场尤其倾向于默契定价，因为在其他人发现之前，通过不声不响的交易中可能能“骗”得的利润是微不足道的。
Now imagine what happens when prices are set by computer software. In principle, the launch of, say, a smartphone app that compares prices at petrol stations ought to be a boon to consumers. It saves them the bother of driving around for the best price. But such an app also makes it easy for retailers to monitor and match each others’ prices. Any one retailer would have little incentive to cut prices, since robo-sellers would respond at once to ensure that any advantage is fleeting. The rapid reaction afforded by algorithmic pricing means sellers can co-ordinate price rises more quickly. Price-bots can test the market, going over many rounds of price changes, without any one supplier being at risk of losing customers. Companies might need only seconds, and not days, to settle on a higher price, note Messrs Ezrachi and Stucke.
现在想象一下：如果计算机软件设定价格时会是什么情况。原则上，推出一个加油站价格比从此的智能手机应用会是消费者的福音。这为省得他们开车寻找最便宜的加油站，但是这样的应用也使零售商可以很容易监控彼此之间的价格。任何一家零售商都不会有降价的动力，因为机器销售员不会错失任何有利商机。算法定价机制造成的快速反应意味着商家会更为快速地相应上调价格。定价机器人可以测试市场，经历不断变化的价格，而供应商不用面临失去客户的风险。 Ezrachi和Stucke指出，企业上调价格的时间可能只需要几秒钟的时间，而不是几天。
Their concerns have empirical backing. In a new paper**, the authors outline three case studies where well-intentioned efforts to help consumers compare prices backfired. In one such instance, the profit margins of petrol stations in Chile rose by 10% following the introduction of a regulation that required pump prices to be displayed promptly on a government website. This case underlines how mindful trustbusters must be about unintended consequences. The legal headache for them in such cases is establishing sinister intent. An algorithm set up to mimic the prices of rival price-bots is carrying out a strategy that any firm might reasonably follow if it wants to survive in a fast-moving market. Online sellers’ growing use of self-teaching algorithms powered by artificial intelligence makes it even harder for trustbusters to point the finger. A cabal of AI-enhanced price-bots might plausibly hatch a method of colluding that even their handlers could not understand, let alone be held fully responsible for.
他们的担忧是有实证依据的。在一篇新发表的论文中，论文作者研究了三个案例。意图帮助消费者比价的行为遭到了反周。在一个例子中，在智利发布一项要求在政府的网站及时显示油价的监管政策之后，加油站的利润率上升了10％。这个案例表明，精明的反垄断者必须考虑到意料之外的后果。在这此案例中，法律破绽是恶意合谋的意图。一个用来模仿竞争对手价格机器人的算法将会实施一种策略，即是一家公司想要在一个快速发展的市场中生存下去，可能会合理地遵循的策略。在线零售商越来越多地使用由人工智能主导的自主学习算法，这就很难让反垄断者有所指责。一群人工智能增强的价格机器可能会巧妙地孵化出一种欺骗手段，即使他们的操作人员也无法理解，更不用说要负全部责任了。
Since legal challenges are tricky, argue Messrs Ezrachi and Stucke, it might be better to direct efforts at finding ways to subvert collusion. Trustbusters could start by testing price-bots in a “collusion incubator” to see how market conditions might be tweaked to make a price-fixing deal less likely or less stable. A “maverick” firm, with different incentives to the incumbents, might have a lasting impact; an algorithm programmed to build market share, for instance, might help break an informal cartel.
由于从法律层面来挑战很棘手，Ezrachi和Stucke认为，更好的办法是找到破坏共谋的方法。 反垄断者可以一开始测试“合谋孵化器”中价格机器人来了解如何调整市场条件，从而使价格固定的交易产生起伏。一个“特立独行”的公司，对职员有不同的激励，可能会产生持久的影响; 例如，一个编制市场份额的算法可能有助于打破一个非正式的企业联合。
Regulators might also explore whether bots that are forced to deal directly with consumers—say, through an app that sends an automatic request to retailers when a petrol tank needs filling—could be enticed to undercut rivals. Or they might test to see if imposing speed limits on responses to changes in rivals’ prices hampers collusion. It may be that batching purchases into bulky orders might thwart a collusive pay-off by making it more profitable for robo-sellers to undercut rivals.
监管机构可能也在探索否存在着一种直接与消费者交易的僵尸程序——比如说，通过一个应用，在油罐需要加油时自动向零售商发送请求——会诱使定价低于竞争对手。或者，他们可能会进行测试，看是否应限制对竞争对手价格变化作出反应以形成默契的速度。有可能以批量采购的订单的方式来阻止有默契地支持，这样就能让机器销售通过比竞争对手的价格更低来实现更多的利润。
Never knowingly undersold
永远不会故意压价
The way online markets work calls for new tools and unfamiliar tactics. But remedies have to be carefully tested and calibrated—a fix for one problem might give rise to new ones. For instance, the more consumers are pushed to deal directly with price-bots (to thwart the transparency that allows rival sellers to collude), the more the algorithms will learn about the characteristics of individual customers. That opens the door to prices tailored to each customer’s willingness to pay, a profitable strategy for sellers.
在线市场的经营方式需要新的工具以及不太为人所知策略。不过，补救措施必须经过仔细的测试和校准——因为解决一个问题可能会导致新的问题的产生。 例如，更多的消费者被迫直接面对价格机器（以阻止允许竞争对手的卖家串通的透明度），那么算法就会更了解个体客户的信息。这打开了为每个客户的意愿支付价格的大门，这是商家的盈利策略。
Still, there is one old-school policy to lean on: merger control. There is growing evidence in old-economy America that trustbusters have been lax in blocking tie-ups between firms. A market with many and diverse competitors, human or algorithmic, is less likely to reach an effortless, cosy consensus about what is the “right” price for sellers, and the wrong price for consumers.
还有一个老套的政策支持：兼并控制。 在美国旧时的经济中，越来越多的证据表明，反垄断者在阻止企业间的合作方面一直处于松懈的状态。一个拥有众多竞争对手的市场，不论是人还是算法，都不太可能就卖家的“正确”价格和消费者的错误价格达成一个轻松，令人舒服的共识。
编译：马佳艺
审校：靳婷
编辑：翻吧君
来源：经济学人