Stay up to Date
Our weekly insiders' guide to political and policy news (Saturdays)
Exclusively for members.
SANDAG is claiming it did not know a forecasting error staffers discovered in 2015 would ultimately lead to voters being offered a false promise in 2016. But the agency’s own staffers made clear to SANDAG executives the two went hand in hand.
Under mounting pressure following revelations SANDAG knowingly misled voters in November, the agency has scrambled to explain itself.
Their primary defense, though, is undone by a single slide – from the agency’s own internal communications.
Last week, Voice of San Diego reported SANDAG’s chief economist and a staffer exchanged panicked emails in fall 2015 after discovering a crucial error in the agency’s economic forecast. The error had significant impact on Measure A, a sales tax proposal SANDAG put before voters in November. The ballot measure said it would raise $18 billion; the flawed forecast meant it would actually raise far less.
Since the investigation was published, SANDAG officials have claimed multiple times they didn’t know those two things were connected – that is, that the forecasting error translated to a false number being put before voters – and therefore the agency did not deliberately lie to the public.
“The eye-catching emails quoted by the press, and attributed to our new chief economist, Ray Major, pertained to income growth, one of numerous factors used to estimate taxable retail sales in the agency’s forecast model,” SANDAG executive director Gary Gallegos wrote in a letter to the Union-Tribune. “Those emails were not related to Measure A.”
But the lengthy presentation that staff economists delivered to their bosses four months before SANDAG’s board voted to put Measure A on the ballot left nothing to the imagination.
SANDAG staffers didn’t force the high-level folks to make the connection between retail sales – the piece of the forecast that contained the error – and tax revenue the agency might get on their own. They made sure the connection was clear.
The basics here are this: SANDAG is claiming it did not know the error it discovered would ultimately lead to voters being offered a false promise. But the agency’s own staffers made clear the two went hand in hand.
Those panicked emails between Major and another SANDAG staffer happened when they realized something big: The agency’s forecast was expecting wages in San Diego to grow at more than double their historic rate, which meant everyone in San Diego would spend wildly more on items subject to sales taxes than they ever had. That’s what led to SANDAG’s $18 billion revenue estimate for Measure A.
Staff members were alarmed by what they had discovered.
They then spent a month preparing a rigorous presentation describing the flaws in the agency’s long-term forecast. The presentation specifically described that a more reasonable forecast would mean the agency would collect much less in tax revenue – both from its existing sales tax and from a new sales tax that would become Measure A.
Gallegos’ note to the U-T that “Those emails were not related to Measure A” is partially true — Major wrote “omg” and “wtf” in emails in response to SANDAG staffer Dmitry Messen, who told him what the forecast expected for income growth. But that happened within a longer conversation that was about both income growth and the growth in taxable sales. In fact, at the start of the conversation, Messen explains that the two are directly related.
“The absolute increase in (taxable retail sales) follows the increase in personal money income,” Messen wrote.
In his letter to the U-T, Gallegos wrote it was not until a year later, in the same month that voters weighed in on Measure A, that SANDAG identified the “root cause” of the faulty forecast. He concedes that the faulty forecast was used to derive Measure A’s revenue total, which was in turn used to develop the list of projects promised to voters if the measure had passed.
The agency’s deputy executive director, Kim Kawada, made a similar claim at last week’s three-day retreat for SANDAG’s board members.
She said the agency knew its forecast was overestimating future wage growth, and therefore how much people in the county would spend on items subject to sales taxes, but it did not know those problems led the agency to overestimate how much tax revenue the agency would bring in.
“We didn’t make that connection at the time until we did a forensic look,” Kawada said.
They are both making slightly different versions of the same argument: We knew there was a problem with certain variables in our forecast, but we did not realize those errors were related to our tax revenue expectations.
Again, these explanations are severely undermined by the presentation SANDAG executives received in which the connection was made explicit. The presentation included a table that spelled out the implications.
The table was even titled “Projected TransNet Revenues.” That means Kawada and Gallegos are arguing they did not know the forecasting error affected tax revenue despite being presented with a slide that directly spelled out how the forecasting error affected tax revenue.
The above slide directly connects the errors in taxable retail sales that SANDAG staffers had found, and how much revenue the agency was expecting to generate from a sales tax.
“SANDAG POF” in the table refers to the most recent revenue expectation for the agency’s existing tax measure. The three scenarios it spells out refer to how much revenue a half-cent sales tax – like TransNet or Measure A – would bring in given different expectations for how much money the typical San Diego resident would spend each year. So, “$14k Scenario” assumes the average resident would spend about $14,000 a year on taxable items. San Diego’s historical average is about $15,000 per year, and has never surpassed $17,749 per year.
“SR13” refers to “Series 13” – which is the name SANDAG has given to its most recent forecast. It’s the forecast the agency used to derive its revenue expectations for Measure A.
Another slide in the presentation likewise shows how much countywide retail sales would decrease if the agency fixed its flawed assumptions for how much money the typical San Diegan would spend each year.
This chart tells a simple story: Here is how taxable sales activity would decrease in the coming years if we fixed our forecast to bring the retail spending expectations in line with historical averages.
When they were preparing for the meeting, Major asked Messen to explain why the agency’s latest forecast was expecting San Diegans to spend so much money.
There’s another piece of SANDAG’s internal jargon in there. “DEFM” is SANDAG’s name for its base forecast – it stands for Demographic and Economic Forecasting Model. “Series 13” is the 13th iteration of DEFM.
That came up earlier last week when Gallegos tried out a related, but slightly different explanation for the scandal.
A letter obtained by Voice of San Diego that was delivered to SANDAG board members said Major had found an error in DEFM – not an error with the agency’s revenue forecast for Measure A.
“Those emails were not related to the Measure A revenue forecast, as some of the press reports have implied,” he wrote. “In fact, Major was not involved in the Measure A revenue forecast.”
Later in the letter, he denied that the presentation delivered to agency executives related to Measure A.
“The meeting in question took place in December 2015, and the topic was updating DEFM,” he wrote.
“There was no connection drawn between that and the bigger picture revenue forecast for TransNet or Measure A.”
That comment isn’t just contradicted by the slide in the presentation that explicitly connects the forecasting error to TransNet revenue. It’s also contradicted by a document SANDAG provided to Voice of San Diego in which the agency explains how it derived its forecast revenue.
It uses DEFM.
The relatively wonky document actually describes a pretty simple process.
First, the agency takes DEFM and extends it to cover the time period covered by the ballot measure. Then, it takes the total taxable sales that are expected by the forecast in that time period and multiplies it by 0.05, the amount of the tax. Then, it subtracts out the amount of money the state takes in administrative costs.
In other words, the idea that an error in DEFM is something separate from an error to Measure A’s revenue forecast doesn’t make any sense. In SANDAG’s own words, it uses DEFM to determine how much a new sales tax can raise.
The other defense SANDAG has used is that staff did not know what part of the forecast was causing wages and taxable sales to grow beyond reasonable expectations. They knew there was a problem, but they didn’t know what was causing it.
“SANDAG staff recognized more than a year ago that the model’s taxable retail sales estimates appeared aggressive, but it was not immediately clear why,” County Supervisor Roberts, who is chairman of the SANDAG board of directors, wrote in a Voice of San Diego op- last month. “Nor was it clear what, if any, impact that might have on the TransNet revenue estimates.”
The presentation and emails make that second sentence demonstrably false. The implications for TransNet revenue were clear, and staff shared them with agency executives.
But the first part is true. In the presentation, staffers put forward a series of theories that might be leading to the errant wage and taxable sales numbers. Eventually, the agency would learn that those theories weren’t right.
The error? It was from a simple copy-paste mistake in the formula.
But Gallegos and Kawada did not tell staff to get to the bottom of the issue following that presentation. Instead, staff was simply told to continue the process of updating the forecast for its next iteration, Series 14.
That instruction changed in November, after Voice of San Diego first published a series of stories revealing that San Diegans would need to spend more than they ever had for Measure A to meet its revenue projection.
Voice of San Diego’s story, however, didn’t include any information that executives hadn’t seen nearly a full year earlier.
The internal revelation didn’t spur executives to get to the root of the problem. That took a public airing of the same information.