Watchdog report also cites progress curbing I.D. theft.
The 2016 filing season brought improvements in the Internal Revenue Services’ level of customer service at call centers, and in efforts to thwart the rising threat of fraud based on taxpayer identity theft, the Government Accountability Office reported.
In a report on a five-year lookback released on Tuesday, auditors found that callers reached a live assistor for tax help 72 percent of the time, compared with only 37 percent in 2015 and seven percentage points higher than predicted. The average wait time was 11 minutes, compared with 18 minutes the year before.
Following years of complaints about budget cuts, Congress in late 2015 gave the agency a funding increase earmarked for improving customer service. As usual, the IRS reduced the level of service before and after the filing season to focus on tasks such as correspondence.
Total call volume rose by 2 percent over 2015’s volume, to more than 114 million calls. Those were handled by a staff increase of 23 percent, which also included 250 full- time employees working on assistance for identify theft victims.
While identity thieves continued to expand methods for scamming people and filing returns using stolen personal information, the number of new ID theft cases dropped in 2016 “because IRS improved its ability to detect fraud before processing the return,” agency officials said.
The report also discussed the agency’s handling of tax return processing, noting some errors in treatment of the Affordable Care Act health insurance premium tax credit.
The agency’s overall performance metrics, GAO noted, are not made easily available to taxpayers. That’s why among the watchdog’s four recommendations is that IRS create an online dashboard to display standards and progress metrics.
IRS officials agreed with most recommendations, but said some of the problems GAO cited are not widespread, and argued with the finding that the extent to which internal control processes prevent the release of fraudulent refunds is unknown.