Ill-gotten gains are artificially bounded by zero, as there is no data on losses experienced by cybercriminals. Thus the shape and skew (mean and stdev too!) of frequency distributions are incorrect. Also: "Cybercrime surveys... exhibit a pattern of enormous, unverified outliers dominating the data... 90% of the estimate [was] from the answers of 1 or 2 individuals." The significant harm experienced by users rather than the small gain achieved by hackers is the true measure of the cost of cybercrime.
Interactive graphic via WSJ: New York City-area cell phone service has a 97% success rate. Yet failed calls in a recent Nielsen test show patterns along roadways and various city neighborhoods and suburban counties. The cause? Unclear.