Sometimes people, when comparing themselves with others, take a host of actions that are destructive to those around them, even when these actions imply self-inflicted costs. "Pulling down" other more successful individuals may have both direct and indirect detrimental effects on productivity and efficiency. On one hand, welfare is reduced directly as output is destroyed, and indirectly if their threat induces ex-ante behavioral responses in the form of lower levels of effort and investment. Consequently, linking reactions to upward social comparisons and their effect on effort levels may help explain the considerable variability in how people have been shown to react to such comparisons. In this paper, I develop a two-stage, two-agent model of strategic behavior that integrates the role of inter-personal comparisons with conventional neoclassical economic preference theory to analyze how interpersonal comparisons lead to destructive behavior and affect levels of effort. The experiment, designed to test the predictions of the model and tease out the mechanisms that drive destructive behavior, builds on the two-stage "money burning" game. The experimental games were carried out in Bolivia among 285 dairy farmers. Results show that people that were above the within-group mean, in average exert less effort when comparing themselves with others (the "guilt" case); while people below the within-group mean exert more effort (the "keep-up-with-the-Joneses" case). People who fear the envy of others decrease their effort exerted, specially if they are highly ranked. Results from the money burning game show that people below the mean took in average more destructive behavior than people above the mean. Of all the participants, 55% took at least one destructive action against somebody in their group reducing their output by 34%. People seem to be averse to disadvantageous inequalities, but not averse to advantageous inequalities. Moreover, people destroy less the bigger the advantageous difference is but destroy more in the oposite case.