腐败犯罪的治理是当前我国犯罪治理中的重大课题。在“行贿受贿并重惩治”的刑事政策指导下，《刑法修正案（九）》对行贿罪宽宥制度做出调整，限缩了特别自首的从宽幅度。分析法条后可以发现，修法后行贿罪特别自首制度的从宽幅度较之于自首、立功并无显著优势，其适用范围也与自首制度高度重合，这说明我国行贿罪宽宥制度已经转型为“弱宽宥模式”。为进一步探究这一立法动向的现实影响，本文进行了相关经验研究。一方面，统计《刑法修正案（九）》修法前后的法检大数据，总结修法前后行贿罪惩治态势的变化；另一方面，以2020年为样本，依托相关平台进行裁判文书检索，对比一定条件下行贿人与受贿人的文书检索成功率。研究结果表明，行贿罪“弱宽宥模式”在司法实务中并未很好地落地。我国对行贿人惩治力度的大幅度增强主要发生在2014年以前，2015年修法后，行贿人年度惩治人数反而产生波动，甚至可以在统计数据中发现“法外施恩”现象存在之端倪。

基于此，有必要对惩治行贿罪的刑事政策导向做梳理与反思。自建国以来，我国对于行贿罪的惩治政策中都暗含着惩治行贿服务于惩治受贿的导向，此即为“重受贿轻行贿”的刑事政策。其主要理论依据为罪责刑相适应原则、立法效果论。“重受贿轻行贿”的问题在于，一旦极端化该理论，可能得出行贿非罪的结论，使得司法办案部门丧失与行贿人的“议价权”。近年来，随着并重惩罚论逐渐发展成熟，刑事政策也越来越体现出“行贿受贿并重惩治”的导向。然而，该理论一方面对于行贿受贿行为的因果关系做了过于机械化的解读，忽略了对国情、社会背景的关照，另一方面期望以严刑重法预防犯罪，这是对刑罚威慑性效用的过分夸大，有重刑主义复刻之嫌。相较之下，行贿罪“强宽宥模式”更能适应我国行贿罪惩治现状。即以特别自首制度为主，以一般宽宥制度为补充，形成层层递进的行贿罪宽宥体系。一方面，该模式强调充分运用不对称博弈的囚徒困境，意在发案前破坏行受贿双方之间的信任关系，发案后攻破双方“攻守同盟”。另一方面，该模式体现了功能主义立法观，即在立法设计中充分考虑我国腐败黑数高的社会现状以及司法办案机关进行辩诉交易的现实需求。至于行贿罪“强宽宥模式”对刑罚特殊预防功能的削弱，只需综合运用罚金刑、非刑罚处罚条款、行贿人黑名单制度进行弥补即可。未来，应当保持开放心态，探索污点证人作证豁免制度，积极进行制度创新。

At present, the governance of corruption is an important subject in Chines crime governance. Under the guidance of the criminal policy of "the equal punishment of bribing and accepting the bribe", The Amendments IX to the Criminal Law has adjusted the leniency system for the crime of bribery and has limited the leniency for special voluntary surrender. After analyzing the statutes, it can be found that, compared with general leniency systems such as surrender and meritorious service, the special surrender of the crime of bribery has no significant advantages in amplitude of lenient penalty, and its scope of application also highly overlaps with the surrender system, which shows that the leniency system for crime of bribery in our country has been transformed into a "weak leniency model". In order to further explore the practical impact of this legislative trend, this paper has conducted relevant empirical research. On the one hand, statistics on the big data of judicial and procuratorial work before and after the Amendment IX to the Criminal Law, then compare the change of relevant statistical data; On the other hand, take the year of 2020 as a sample, through relevant platforms to carry out judgment document retrieval, then compare the success rate of document retrieval between the briber and the bribe accepters under certain conditions. The results show that the "weak leniency model" for the crime of bribery has not landed well in judicial practice. In our country, the substantially increase of the punishment of bribers actually occurred before 2014, but after the Amendment IX to the Crime Law in 2015, the number of bribers punished has fluctuated, and we can even find the clues of the "extra-legal leniency" phenomenon in the statistical data.

Therefore, it is necessary to sort out and reflect on the criminal policy of punishing bribe crime. Since the founding of our country, the punishment of bribe crime has always implied the idea of punishing the crime of bribery in the service of punishing the crime of accepting bribes, that is, the criminal policy of “the punishment of bribing is lighter than that of accepting the bribe”. Its main theoretical basis is the principle of suiting punishment to crime and the theory of legislative effect. The problem of "the punishment of bribing is lighter than that of accepting the bribe" is that, once the theory is taken to extremes, it may conclude that the crime of offering bribes is not guilty, which makes the judicial department lose the "bargaining power" with the briber. In recent years, the theory of “the equal punishment of bribing and accepting the bribe” has become more and more mature, this orientation is also increasingly reflected in criminal policies. On the one hand, the theory has made an over-mechanistic interpretation of the causality of bribing and accepting bribes, ignoring the national conditions and social background. On the other hand, it expects to use severe punishment to prevent crime, which is an over-exaggeration of the deterrent effect of punishment. In contrast, leniency system for bribery crime should be "strong leniency model", that is, the special voluntary surrender system is the main, the general leniency system is the supplement, and forms a progressive leniency system. On the one hand, this model makes full use of prisoner's dilemma of asymmetric game, aiming to destroy the trust relationship between the bribery parties before the incident and reduce the difficulty of obtaining confessions after the incident. On the other hand, this model reflects the functionalism legislation, that is, in the design of legislation, we should take into full account of the high black number of corruptions in China and the case-handling authorities’ practical needs of plea bargaining. As for the problem that the "strong leniency model" weakens the special preventive function of the penalty, it is enough to make up by comprehensively using fines, non-penal punishment clauses, and the blacklist system for briber. In the future, we should keep an open mind, explore the exemption system for stain witness, and actively carry out system innovation.