Fully progressive taxation
Fully progressive taxation
复制标题
完全累进税制
DOI：
10.1016/0047-2727(80)90010-9
复制
发表时间：
1980
影响因子：
9.8
通讯作者：
P. Hammond
中科院分区：
文献类型：
--
作者：
P. Dasgupta;P. Hammond
Mirrlees(1971)理所当然地成为所得税理论探索的标准参考文章。然而，正如他在那篇论文的最后一段中承认的那样，所得税并不能真正充分利用那里经常可以获得的所有信息。碰巧的是，我们通常不仅可以观察到每个工人的收入，还可以观察到他的工作时数。在这种情况下，可以推断工人的技能，因为它(在Mirrlees的模型中)正好与他每小时工作赚取的收入成比例。如果可以观察到技能，人们会问为什么不应该直接征税？在这种情况下，似乎最优的税收方案将是对技能一次性征税，使第一最优方案得以实现--至少在工人都有相同品味的情况下是这样，就像他们在米尔利的模型中所做的那样。事实上，Mirrlees确实暗示这是一种可能的结果。在这篇文章中，我们根据我们其中一人最近从事的关于激励相容的一些工作来重新审视这一论点[Hammond(1979)]。我们的结论是，如果工人的技能水平不是直接知道的，而是必须从他每小时的工资收入中推断出来的，那么通常不可能达到第一最优。正如Mirrlees(1971，p.205)和更广泛的(1974，p.251)，(1977，定理3.1)所表明的那样，首先最好的效用最优通常涉及到技能较高的工人获得的效用水平低于那些被赋予较低技能的工人。当休闲是一种正常的好处时，情况肯定是如此。因此，在试图实现优先最优时，我们面临这样一个事实，即工人将选择隐藏他们的技能。那么，似乎我们在不了解工人技能的情况下所能做的最好的事情
Mirrlees (1971) has justly become the standard reference article for theoretical explorations in income taxation. Yet, as he acknowledges in the last paragraph of that paper, an income tax does not really make the most of all the information there is often available. It so happens that we can usually observe not only each worker’s income, but also the number of hours he works. In this case, it is possible to infer the worker’s skill, since it is (in Mirrlees’ model) just proportional to the income he earns per hour of work. If skill can be observed, one asks why it should not be taxed directly? In which case, it might seem, an optimal scheme of taxation would be a lumpsum tax on skill, enabling the first-best optimum to be reached-at least in the case where workers all have the same tastes, as they do in Mirrlees’ model. In fact, Mirrlees does suggest this is a possible outcome. In this paper we re-examine this argument in the light of some work on incentive compatibility, which one of us has recently undertaken[Hammond (1979)]. We conclude that, where a worker’s skill level is not known directly, but has to be inferred from his wage income per hour of work, it is usually not possible to reach a first-best optimum. As Mirrlees (1971, p. 205) and, more generally (1974, p. 251),(1977, theorem 3.1) has shown, a first-best utilitarian optimum would usually involve the more skilled workers attaining a lower utility level than those endowed with less skill. This is certainly so when leisure is a normal good. Thus, in trying to implement the first-best optimum, we face the fact that workers will choose to hide their skill. It seems, then, that the best we can do without knowing workers’ skills in