Paradox of Self-Amendment by Peter Suber O. Tax reflexivities O.1. Pyramiding Taxes. If an employer pays an employee's income tax as an extra benefit, then the amount of the tax paid is considered to be extra income to the employee. Because it is income, it is taxable. If the employer pays the tax due on that extra increment of income, and then on the next increment due on her second payment, and so on, then an infinite series is created. However, the series converges and the exact amount of tax due can be calculated, which is always more than the employee would have to pay if she paid her own taxes directly. Hence this perquisite is rarely used. Whether taxes pyramid in this way is a matter of policy. They are allowed to do so in the case of employers and employees. But to encourage leases in which the lessee pays the lessor's taxes on the leased property, §110 of the Internal Revenue Code (IRC) stops the pyramiding by stipulating that payment of the lessor's taxes is not considered income to the lessor. Pyramiding down and up in oscillation would be possible if the late-filing charge or other penalties could be deducted in an amended return. The larger deduction than originally taken would decrease the tax due, decreasing the penalty, decreasing the allowable deduction, increasing the tax due, and so on in a series that would probably always converge. IRC §162 (f) prevents the deduction of such penalties.[Note 36] O.2. Mutually Deferring Provisions. Under IRC §1015 (a) the taxpayer's basis on the sale of a gift, if a gain, is the donor's basis, and if a loss, is either the donor's basis or the fair market value at the time of the gift, whichever is lower. Suppose the donor buys something for $200 (donor's basis), gives it to the donee when it is worth $100 (fair market value), and the donee sells it for $150. Then the sale is both a gain and a loss. If the sale is considered a gain, then we use the donor's basis ($200), yielding a $50 loss. If the sale is considered a loss, then we use the lower of the two bases ($100), which yields a $50 gain. If the sale is a gain, then it is a loss; and if it is a loss, then it is a gain. The Code does not solve this problem, but Treasury Department Regulations do. Regulation §1.1015-1 (a) cuts the Gordian knot by allowing no deduction and imposing no tax on such transactions, making it immaterial whether they were gains or losses. O.3. Imputed Income. If one performs a service with a market value for oneself, one may be said to have "earned" the value of the service as income. Such "earnings" are called imputed income, and are traditionally not taxed. For example, a housepainter who paints her own house is not taxed on the value of her work, nor is the homeowner who lives in her own house taxed on the imputed rent. But an insurance agent was once taxed on "commissions" that he "earned" on policies that he bought from his company for himself. Commissioner v. Minzer, 179 F.2d 338 (1960). Minzer claimed that the "commissions" were subtracted from the purchase price, giving him a discount and no income. The IRS claimed that he performed a service for compensation. Minzer paid the price of his policies without the commission, whereas other buyers pay both the price and the commission. Because no commission was paid, none could have been received as income in the normal sense. But just because none was paid, one was "earned" as imputed income —a benefit with a cash value in the market place was received in exchange for labor. Because Minzer was dealing with himself, he earned a commission because he paid none, and he was taxed on it. Other questions about self-dealing from outside tax law: Why are contracts with oneself void? In a slave-state where slaves were chattels, could a slave take title to herself? P. Circular reasoning Judges and lawyers are not immune to the fallacies of circular reasoning, and many types frequently occur. But what is more interesting is circular reasoning that is permitted or required by law, without regard to its logical validity. We've seen many cases under other heads above; here are some others. P.1. Mutual Corroboration. In order to minimize the risk of collusion in divorce cases in fault jurisdictions, many states have statutes that require that a spouse's fault be shown with evidence other than her admission of fault. In Husband v. Wife, 253 A.2d 63 (1966) the wife denied committing adultery or confessing to adultery to her husband, but he presented circumstantial evidence that she had both committed adultery and confessed it to him. The circumstantial evidence of the wife's alleged confession was legally insufficient taken alone, but the court reasoned that the circumstantial evidence corroborated the husband's in-court testimony, which in turn corroborated the circumstantial evidence. Together these mutually corroborating stories showed (the court held) that the wife's denial was perjured, and hence that she was guilty by a sufficiency of evidence. The court unabashedly concluded, after a string of citations, that the wife's "perjury impels acceptance of the husband's testimony concerning her admissions. We accordingly have a situation in which the indisputable circumstantial evidence corroborates the husband's testimony, which in turn corroborates the circumstantial evidence." Whether the statute designed to minimize collusion required this result is one question; another is whether the court should have inferred something from the fact of the dispute, namely, that a contested divorce does not present a large risk of collusion. A generalized version of the problem above occurs whenever a mass of evidence is found sufficient when the bits comprising the mass are separately insufficient as a matter of law. The support that each bit provides for and receives from the others, by completing a coherent picture, is commonly and probably harmlessly found to be persuasive. But logically it is difficult to distinguish from the mutual proof of two uncertain propositions. See Peters v. U.S., 408 F.2d 719 (1969), dissent: "Adding hearsay to hearsay [even in administrative hearings where hearsay is admissible] is like adding zero to zero which still equals zero." 144