(Mark Lennihan/AP)

The AT&T-Time Warner merger could end up costing consumers less money than what some earlier estimates suggested, the government's star witness said in federal court Tuesday as he clashed repeatedly with company lawyers over key figures in his economic analysis of the deal.

Instead of paying a minimum of 27 cents more per month on their bills as a result of the deal, TV subscribers could conceivably pay a smaller premium of at least 13 cents a month more — a downward revision in the projections of Carl Shapiro, an economist at the University of California at Berkeley. Toward the other end of the range, the government has said consumers could collectively pay hundreds of millions of dollars a year, or 45 cents more per month on each of their bills.

Shapiro's concession could weaken the government's case to block the $85 billion merger, which seeks to meld AT&T's wireless distribution capability with Time Warner's massive library of TV and film content. AT&T has sought to demonstrate that the deal will not lead to price increases and has argued that, in fact, it will lead to price decreases for consumers.

Contributing to the revised numbers was newer profit margin data from AT&T that Shapiro said was not available to him when he completed his initial report. He also said he had “made a mistake” in his earlier testimony when he said that certain numbers in a report from Boston-based consulting firm Altman Vilandrie had not changed pertaining to the number of video subscribers who would probably switch TV providers in the event of a protracted programming dispute involving Time Warner and those distributors. Those numbers had changed from one version of the report to the next, and Shapiro said he was not aware of the change.

The heated exchange between Shapiro and Daniel Petrocelli, the lead attorney for AT&T and Time Warner, grew so tense that at one point U.S. District Judge Richard Leon intervened to prevent the two men from talking over each other.

“Let me finish my answer,” Shapiro said to Petrocelli as Leon sought to keep the proceedings on track. “It is not my practice, sir, to take this [newer] data without checking it out. "

“You're sticking to your number because it's a higher number,” Petrocelli alleged.

“That's completely false,” Shapiro said. “AT&T for months didn't provide the data.”

Shapiro said he was simply trying to be fair to AT&T by providing a more conservative lower bound in his range of predicted price effects and complained that he was being punished for being cautious.

“That was your choice,” Petrocelli said.

Petrocelli took aim at other parts of Shapiro's analysis, accusing him of omitting from his report a steep decline in late 2016 of video subscribers from the cable and satellite TV industry that could undercut his analysis.

Shapiro, meanwhile, argued that AT&T's defense of the deal turns on a narrow understanding of consumer benefits. Although AT&T has promised price cuts for customers of its own TV service, the deal will still result in price increases for customers of other cable and satellite companies, said Shapiro. In his analysis, AT&T will pass along roughly a fifth of its merger-related savings to consumers, while other TV providers will pass along roughly two-thirds of the added costs they incur from the deal.

Shapiro added that he wouldn't be testifying if he were not confident in the quality of his economic modeling.

“In the end,” said Shapiro, “What are the reliable numbers? What are the best estimates? That's why I'm here.”

Shapiro may return to the stand on Thursday afternoon for more questioning, if attorneys for both sides determine it is necessary. Closing arguments in the trial are expected next week.

Correction: An earlier version of this article said Shapiro admitted he made a mistake in his analysis. In fact, he was saying he made a mistake in his testimony in describing the study from Altman Vilandrie.