Using a natural experiment in the taxicab industry to analyze the effects of third-party income reporting