High School Research

 How Much Would A Wealth Tax Reduce Inequality In The United States?

Abstract

It is widely accepted that implementing a wealth tax would increase the progressivity of the U.S.’ current tax system. The leading U.S. wealth tax proposal is Elizabeth Warren’s “Ultra Millionaire Wealth Tax”, which annually taxes two percent of wealth over $50 million, plus an additional four percent on wealth above $1 billion. In this paper, I examine whether a wealth tax would decrease wealth inequality by more than an equally-sized income, capital gains, or estate tax (scaled by attempted revenue collection). I analyze two factors: the current distribution of wealth that would be subject to the tax, and the potential for evasion of each tax. I conclude that, due to the significantly higher rate of evasion, the wealth tax yields fewer funds for redistribution, and thus is worse at decreasing inequality than an equally-sized income, capital gains, or estate tax despite its wealthier tax base.

Gila Koren, 2023. How Much Would A Wealth Tax Reduce Inequality In The United States? 

Using Novel Searching Strategies To Deal With Liars in The Real World.

Abstract

This paper tackles the question of obtaining information while dealing with dishonest subjects. This paper models this interaction through the common higher or lower searching game, where one player, “the chooser,” picks a number from a set of Z integers. The other player, “the guesser,” attempts to minimize his expected value for the number of guesses it takes to arrive at “the chooser’s” number, with only the feedback of higher or lower after each guess. In this variation of the game, “the chooser” randomly tells the truth about whether his number is higher or lower h% of the time, where h ∈ [50, 100]. I tackle this question with two methods: a modified binary search algorithm and an information-based approach. I first explore the modified binary search algorithm, where “the guesser” repeats guesses based on a certainty threshold, γ, which represents their level of skepticism. I find that more trusting people (with a higher tolerance threshold) perform better with more honest people, and more skeptical people perform better with increasingly dishonest people. I then look into the information-based approach, where “the guesser” guesses whichever number maximizes his information about which integer the “the chooser” chose. I find that my information-based approach outperforms the binary search-derived method, revealing that new methods need to be developed, with liars in mind, in order to best deal with lies in the real world.

Gila Koren, 2023. Using Novel Searching Strategies To Deal With Liars in The Real World.