演算法之競爭法議題—以聯合行為為中心

近幾年來，事業使用演算法作為其訂價工具愈來愈普遍，而演算法的出現，大幅提升了市場透明度及事業的互動頻率，其對於市場結構之改變不可小覷。而這樣的改變，使得明示共謀變得更容易形成，此部分已有實際案例發生；演算法也可能使默示共謀更容易發生，然目前尚未有實際案例可佐證，也因此仍有認為，演算法不一定會使默示共謀更容易發生，反而有可能使默示共謀更難以形成之爭論。在演算法促成明示共謀之部分，其對於競爭法造成的影響在於，其大幅減少人類於共謀運作當中所需要的溝通，使明示共謀之證據更加難以尋得。對此，本文認為演算法所帶來舉證上的困難，於強調溝通性證據之重要性的國家會造成較大影響，然觀諸我國公平會處分及法院判決，在使用間接證據推論合意時並未特別強調溝通性證據，時常僅使用經濟性證據即推定合意，若實務之操作方式未改變，演算法對於我國執法機關的舉證並不會帶來太大的影響。至於演算法可能促成默示共謀的部分，由於過往默示共謀需要有特定條件（通常為寡占市場）較為容易發生，而要滿足這些條件並不容易；再加上默示共謀為處於寡占市場中之事業為極大化自身利益所不得不為，因此為公平法所不罰。然演算法的出現，使得默示共謀可能發生於寡占市場以外之市場，使其發生頻率變得頻繁，對於市場競爭產生不良影響，因而引發應不應修法處罰默示共謀之討論。本文認為在尚未確定演算法造成的影響前，不應輕舉妄動，以免使無辜事業受罰。另外，本文也介紹了目前各國面對演算法對於競爭法帶來的衝擊所持的態度，以及討論了公平法第14條第3項、公平法第15條、公平法第20條第4款以及公平法第25條是否可用來處理演算法所造成之競爭法議題；並認為價格管制不可行、以及設計遵守原則有其困難之處。期許能使執法機關在面對演算法議題時作為參考。

關鍵字

演算法；共謀；聯合行為；默示共謀；間接證據；市場透明度

並列摘要

In recent years, it has become more and more common for enterprises to use algorithms as their pricing tools. The emergence of algorithms has greatly improved the transparency of the market and the interaction frequency of enterprises, and its impact on the change of market structure can not be underestimated. Such changes make it easier to form explicit collusion, which has already happened in actual cases. Algorithms may also make tacit collusion more likely to occur, but there are no actual cases to support it. Because it has not yet occurred, there are still arguments that algorithms may not make tacit collusion easier to occur, but may make tacit collusion more difficult to form. In the part of the algorithm that facilitates explicit collusion, its impact on competition law is that it greatly reduces the communication required by humans in the operation of collusion, making it more difficult to find evidence of explicit collusion. In this regard, this article believes that the difficulty of proof brought by the algorithm will have a greater impact in countries that emphasize the importance of communication evidence. However, observing Fair Trade Commission decisions and judgments in Taiwan, they don't pay special attention to the communication evidence when they use indirect evidence to infer agreement. They often only use economic evidence to infer agreement. If their operation mode remains unchanged, the algorithm will not have much impact on the proof of our law enforcement agencies. As for the part of the algorithm that may lead to tacit collusion, since tacit collusion in the past required certain conditions (usually in oligopoly market) to occur more easily and it was not easy to meet those conditions, it was not punished by the competition law. However, the emergence of algorithms makes it possible for tacit collusion to occur in markets other than the oligopoly market, making its frequency more frequent and having a negative impact on market competition, thus triggering discussions on whether the law should be amended to punish tacit collusion. This article believes that before the impact of the algorithm has been determined, one should not act rashly, lest innocent businesses be punished. In addition, this paper also introduces the current attitudes of various countries towards the impact of algorithms on competition law, and discusses whether Article 14 (3), Article 15, Article 20(4) and Article 25 of competiton law can be used to deal with competition law issues caused by algorithms. This paper also points out that maximum price regulation is not feasible, and compliance by design has its difficulties. It is expected that the law enforcement authorities can use it as a reference in the face of algorithmic issues.