Monetary rewards granted on a per-publication basis to individual authors are an important policy instrument to stimulate scientific research. An inconsistent feature of many article reward schemes is that they use journal-level citation metrics. In this paper we assess the actual article-level citation impact of about 10,000 articles whose authors received financial rewards within the Romanian Program for Rewarding Research Results (PR3), an exemplary money-per-publication program that uses journal metrics to allocate rewards. We present PR3, offer a comprehensive empirical analysis of its results and a scientometric critique of its methodology. We first use a reference dataset of 1.9 million articles to compare the impact of each rewarded article from five consecutive PR3 editions to the impact of all the other articles published in the same journal and year. To determine the wider global impact of PR3 papers we then further benchmark their citation performance against the worldwide field baselines and percentile rank classes from the Clarivate Analytics Essential Science Indicators. We find that within their journals PR3 articles span the full range of citation impact almost uniformly. In the larger context of global broad fields of science almost two thirds of the rewarded papers are below the world average in their field and more than a third lie below the world median. Although desired by policymakers to exemplify excellence many PR3 articles are characterized by a rather commonplace individual citation performance and have not achieved the impact presumed and rewarded after publication based on journal metrics. Furthermore, identical rewards have been offered to articles with markedly different impact. Direct monetary incentives for articles may support productivity but they cannot guarantee impact.

每次发表授予个人作者的金钱奖励是刺激科学研究的重要政策工具。许多文章奖励计划的不一致之处在于它们使用期刊级别的引用指标。在本文中，我们评估了实际的文章级引文影响约10,000篇文章的作者在罗马尼亚奖励研究结果计划（PR3）中获得了经济奖励，这是一种示例性的按出版物计价的计划，该计划使用期刊指标来分配奖励。我们介绍PR3，对其结果进行全面的实证分析，并对方法论进行科学计量学评论。我们首先使用190万篇文章的参考数据集，将来自五个连续PR3版本的每篇获奖文章的影响与同一期刊和同一年发表的所有其他文章的影响进行比较。为了确定PR3论文对全球的广泛影响，我们随后根据Clarivate Analytics Essential Science Indicators的全球实地基准和百分位等级对它们的引文表现进行了基准测试。我们发现他们的日记中PR3文章几乎均匀地涵盖了所有引文影响。在更大范围的全球科学领域中，近三分之二的获奖论文均低于其领域的世界平均水平，而三分之一以上的论文低于世界中位数。尽管决策者希望表现出卓越的表现，但许多PR3文章的特点是具有相当普遍的个人引文表现，并且未达到基于期刊指标发表后推测和奖励的影响。此外，已经对具有明显不同影响的物品提供了相同的奖励。对物品的直接货币奖励可能会提高生产率，但不能保证产生影响。