Last week TPR shared a new report from Embarcadero Institute disputing the accuracy of the state's methodology for calculating local housing needs. With the state moving to hold local governments accountable for meeting housing production goals and the report finding a 900,000 unit discrepancy, offered here is Embarcadero Institute's response to criticism received regarding the report's conclusions. Report author, Gab Layton, has a Ph.D. in Aerospace Engineering and currently serves on the advisory board of the Stanford Institute for Economic Policy Research. She asserts her report findings are basic arithmetic analyzing data reported by government agencies. All data and analyses are available in spreadsheets published alongside the report so that all work can be reviewed.

Our most recent report, Double Counting in the Latest Housing Needs Assessment, generated a little flurry of activity on Twitter. We thought we'd share the criticisms that surfaced and our response to those criticisms through our newsletter. Our subscriber base includes a large number of city councilmembers, city managers, city planners, and city attorneys across California, many of whom are too busy navigating the COVID crisis to take note of social media.

Professor Chris Elmendorf, UC Davis School of Law, tweeted:

EI (Embarcadero Institute) contrasts the “SB828 double count” with what it calls a “conventional economist approach” under which housing need is equal to projected household growth plus a small vacancy adjustment. This is nonsense. No “conventional economist” would equate California’s housing need with projected household growth. That forecast “bakes in the housing crisis.”

Embarcadero Institute response:

Agreed. A “conventional economist” would not equate California’s housing need with projected household growth. We think Professor Elmendorf may have misunderstood our findings.

The model the Embarcadero Institute refers to as the “conventional economist approach” is the longstanding model employed by the California Dept. of Housing and Community Development (HCD) for the 3rd, 4th, and 5th housing cycles that have projected California’s housing needs since 1998 (and perhaps earlier but we only have information on projections since then).

Leading economists working for the state developed the model. Elmendorf may have misunderstood this history, when we used a term the general public could grasp — “conventional economist approach” — distinguishing HCD’s long-time model from their new 2020 model influenced by Senate Bill 828.

The point made in our report  Double Counting in the Latest Housing Needs Assessment (page 3) was that SB-828’s authors (State Sen. Wiener, the Bay Area Council, and the Silicon Valley Leadership Group) mistakenly assumed that HCD had been calculating housing needs based on the limited equation below (which, as Prof. Elmendorf points out, is not correct): California housing needs = Projected housing needs

But in fact, past models over the decades show that HCD has always calculated California’s housing need = Existing housing need + Projected housing need

We think the authors of SB-828 mistakenly concluded HCD had used the former equation, and probably came to that conclusion because, at the time of the 5th cycle assessments, vacancy rates were unusually high (higher than the healthy benchmark).  As such, the 5th cycle assessments didn’t refer to an ‘existing housing need’ but instead referred to an ‘adjustment for existing excess vacant units.’

In short, we agree that it would be a mistake to equate California’s housing need with projected household growth, and the HCD does not appear to have ever done that.

Professor Elmendorf tweeted:

Yet nothing in the bill as enacted *disallows* HCD from applying a 5% target to owner-occupied housing. Doing so is entirely appropriate where prices have escalated such that the housing market is no longer "healthy"

Embarcadero Institute response:

The exact language of the code that pertains to vacancy rates:

Govt Code 65584.01

For the fourth and subsequent revision of the housing element pursuant to Section 65588, the department, in consultation with each council of governments, where applicable, shall determine the existing and projected need for housing for each region in the following manner:

At least 26 months prior to the scheduled revision pursuant to Section 65588 and prior to developing the existing and projected housing need for a region, the department shall meet and consult with the council of governments regarding the assumptions and methodology to be used by the department to determine the region’s housing needs. The council of governments shall provide data assumptions from the council’s projections, including, if available, the following data for the region:

The vacancy rates in existing housing stock, and the vacancy rates for healthy housing market functioning and regional mobility, as well as housing replacement needs. For purposes of this subparagraph, the vacancy rate for a healthy rental housing market shall be considered no less than 5 percent.

Our point is that the code only stipulates that a “healthy rental housing market shall be considered no less than 5%.” It does not stipulate that owner-occupied housing also be held to this standard. Based on national data, the long-term owner-occupied housing vacancy is 1.5%, and even at the height of the mortgage crisis it barely reached 3%. Knowing also that 5% is outside the HCD's own stated healthy range for owner-occupied housing, it would seem a 5% vacancy rate in owner-occupied housing from most perspectives is not appropriate.

In our next newsletter, we will evaluate a range of housing needs assessment models to see what the experts consider "appropriate".

Professor Elmendorf critique:

Elmendorf argues that the Dept. of Housing and Community Development (HCD) didn’t go far enough. Elmendorf claims in his own report published on UCLA’s website, https://www.lewis.ucla.edu/research/new-approach-housing-element-update/ that the HCD underestimated the housing need for the Bay Area. Elmendorf references Govt Code 65584.01(b)(1)(G), 655841(b)(1),  and 65584.1(b)(2) to make his point. He claims that the HCD:

Ignored a statutory factor (the jobs/housing imbalance) in its determination of housing needs.

Should have applied the cost-burdening adjustment not only to future households but also to existing households.

His own model makes those two adjustments and, citing its outcome, he claims the Bay Area’s 6th cycle target for housing needs should be 686K not 441K housing units.

Embarcadero Institute response:

With respect to the jobs/housing imbalance as a statutory factor (1)

On page 11 of Elmendorf's report, he cites Gov't. Code 65584.1(b) as evidence of HCD's failure to account for the jobs/housing imbalance. That section of the code pertains to fees Councils of Governments may charge cities. We suspect Elmendorf meant to refer to Gov't Code 65584.01(b)(1)(G). This section of the code states that the HCD must "meet and consult" with the Councils of Governments "regarding the assumptions and methodology to be used by the department to determine the region’s housing needs." It also states that the Councils of Government must provide data to the HCD on “the relationship between jobs and housing, including any imbalance between jobs and housing.”

However, the code does not appear to compel an adjustment based on jobs/housing imbalance. Historically, the jobs/housing imbalance has been used by the Councils of Governments to determine the relative share of housing accorded to cities within their region. Separately in an annual housing report produced by the Joint Center for Housing Studies of Harvard University, The State of the Nation’s Housing 2019, claims:

“Over the long run, residential construction should exceed household growth to provide some margin for replacement of older units, demand for second homes, geographic shifts in the population, and a normal amount of vacancies.”

This, in a nutshell, is the approach traditionally used by California’s Dept. of Finance (DOF) and Dept. of Housing and Community Development (HCD) for earlier housing assessments. The only difference being that the DOF separates out seasonal homes (second homes) from their forecasts. Their projected housing needs do, however, factor in household growth, replacement of older units, geographic shifts in the population (net migration), and a normal amount of vacancies (benchmark vacancy rate). Harvard’s list of factors to consider in projecting housing needs does not suggest a jobs/housing imbalance adjustment should be made over and above these factors.

With respect to Elmendorf's cost-burdening adjustment (2)

Elmendorf references Gov't Code 655841(b)(1) on page 8 of his report when discussing the legislative implications for cost-burdening adjustments. We were unable to find that part of the code and suspect he meant Gov't Code 65584.01(b)(1)(H) which speaks to regional cost-burdening data that Councils of Governments must supply the HCD. The Harvard report uses cost-burdening statistics as an indicator of the type of housing that is needed, not the amount of housing that is needed. Nationally, Harvard has found that rising levels of cost-burdening in lower-income households are due to building trends, which primarily serve the higher end of the housing market. From their report:

“The National Multifamily Housing Council also notes that new construction has not even served the middle of the market, with the share of new apartments affordable to median-income renter households dropping to less than 3 percent annually over the last decade. The focus of new construction on higher-cost units has thus shifted the overall distribution of rents upward.” (p.30)

Like Professor Elmendorf, we noted in our report that the HCD cost-burdening adjustment had only been applied to future housing. Since his rebuttal, we have applied Elmendorf’s cost-burdening approach to the SCAG region (see “Six Counties of SoCal” tab on our spreadsheet). This adjustment increased the housing target for the six counties of SoCal overseen by SCAG from 1.34M to 1.94M housing units in the next 8 years.

According to the Harvard report referenced above, 1.18 million housing units were completed across the entire country in 2018. Southern California would have to issue permits for 20% of the U.S housing starts eight years in a row to reach its target under Elmendorf's approach. It’s the equivalent of adding the state of Kentucky to Southern California in that 8-year period. At peak production, California achieved just over 200,000 housing units a year statewide. Southern California alone would have to build 240,000 units a year under Elmendorf’s model. In housing units, this would mean the SCAG region would have to add the housing equivalent of Las Vegas every year. Note that these numbers do not include Elmendorf's second factor - the jobs/housing adjustment.

From the Harvard report:

Professor Elmendorf tweeted:

DOF inferred a “normal” headship rate by averaging the 2000 and 2010 Census. It goes w/o saying that CA’s housing market was not healthy in those years.

Embarcadero Institute response:

We agree the 2010 headship rate might have been affected by the Great Recession, but the Harvard report suggests that household growth is now stabilizing at early 2000 levels (see below). So perhaps 2000 is not an unreasonable benchmark. The DOF did not publish their actual headship rates. They just noted the adjustment had been made with overcrowding and cost-burdening in mind. We will certainly be looking at other models in our next newsletter - including one that uses historical benchmarks 1970 to 2000. We’ll see how they compare.

Our larger point about double counting was that SB-828 did not mention the DOF adjustment for overcrowding and cost-burdening. It wasn’t clear anyone drafting the legislation realized that the DOF had already made some accommodation for overcrowding and cost-burdening. Any adjustment for these factors should be done just one time in the model. Even if Elmendorf believes the DOF adjustment isn’t enough, it doesn’t make logical sense to have two separate adjustments for the same variable in one model.

Professor Elmendorf tweeted:

Setting housing targets is not just math. It's political.

Embarcadero Institute response:

We agree the process is political but since  Cities and counties are to be held accountable by the state for reaching these targets. State funding is tied to these targets

The targets will be used to measure progress in the housing crisis shouldn’t the targets be based on some defensible rationale? Cities and countries around the world face the challenge of anticipating housing needs. Surely there must be some commonly accepted practices and methodologies as Harvard’s The State of the Nation’s Housing annual report suggests. Without a transparent, consistent, logical approach, the public cannot play its crucial role and the state may inadvertently expose itself to legal challenges.

Professor Elmendorf critique on Twitter:

You know it’s extra strong bunk when they can’t even find a pseudo economist who will put their name on the report.

Embarcadero Institute response:

An oversight on our part. The author is Gab Layton. She currently serves on the advisory board of the Stanford Institute for Economic Policy Research. She has a Ph.D. in Aerospace Engineering, a Bachelor of Engineering (Mechanical), worked in scramjet engine research, and at McKinsey & Company. She makes no claim to being an economist — pseudo or otherwise.

The models presented in the report are not the Institute's, they belong to the Dept. of Housing and Community Development. The data analyzed is publicly available and is prepared and used by various government departments. Sources include the American Community Survey, Comprehensive Housing Affordability Strategy Data CHAS (2012-2016), Dept. of Housing and Community Development progress reports, and the Dept. of Finance projection tables.

The report simply analyzes data reported by these government agencies. All data and analyses are available in spreadsheets published alongside the report so that all work can be reviewed. It’s all basic arithmetic. We can assure you that no differential equations were harmed in the making of this report.

Mayor of Palo Alto Adrian Fine; and Legislative Director of California YIMBY, Louis Mirante:

Both men raised questions about the Embarcadero Institute’s funding.

Embarcadero Institute response:

Per publicly available data on Open 990: The Embarcadero Institute has been a 501C-3 since May 2015. Total expenses in 2018 were $104,779. The total revenue in 2018 was $110,000.

Per publicly available data on Open 990: California YIMBY, Louis Mirante's employer, has been a 501C3 since 2017. Total expenses in 2018 were $1,040,826. Total revenue in 2018 was $3,444,880.

Mayor of Palo Alto Adrian Fine tweeted:

The Institute is backed by anti-housing ‘residentialists’ who have donated tens of thousands to local anti-housing/change candidates for council.

Embarcadero Institute response:

We believe he was referring to Gab Layton who is a major donor to those seeking office in California. Her donation history is publicly available through CalAccess https://cal-access.sos.ca.gov/. Since 2015 she has donated to the campaigns of many elected officials at the state and local level on both sides of the housing debate including State Sen. Wiener. She did not support Adrian Fine because she felt he favored an oversupply of office space at the expense of housing needs.

Associate Professor Paavo Monkkonen, UCLA Luskin School of Public Affairs, Director of Latin American Cities Initiative tweeted:

"too boring"

Embarcadero Institute response:

We agree. It’s not the most interesting math, but someone has to do it.

For more detail about the housing assessment errors click here, and for the spreadsheet that outlines all the calculations click here.

This story was republished with permission from The Planning Report.