Can't Fix What You Can't Measure: Correcting the Record on Double Counting in State RHNA Numbers 

The op-eds, published in November by the California Planning & Development Report (CP&DR) "How Much Housing Does California Need?” by William Fulton; and “No Matter How You Calculate It, We Need a Lot of Housing,” by Josh Stephens included misunderstandings addressed below.

1. The Double-Count in the state RHNA numbers -- Step by Step

Fulton in his op-ed correctly states that Embarcadero Institute’s numbers rest on two arguments:

1. The state inaccurately applied a rental vacancy benchmark to owner-occupied housing 

2. The state double-counted overcrowding and cost-burdening

While Fulton agrees with our analysis regarding the state’s inaccurate use of vacancy rates, he focuses his critique on our assessment of the much larger error by the state’s Dept. of Housing and Community Development (HCD ) -- its double-counting. 

To address this critique it’s important to understand that the state’s Regional Housing Needs Assessment (RHNA) has two components:

1. Existing housing need  = the backlog of unmet need.
2. Future housing need  =  housing that will accommodate expected growth in the region between the start and end of the housing cycle (in this case, from 2021 to 2030).

Fulton mistakenly believed that before the passage of SB-828 (Wiener) HCD applied overcrowding and cost-burdening adjustments, but only to future housing needs. Fulton further compounds this misunderstanding by thinking that, after the passage of SB-828, HCD decided to extend the adjustments to all housing -- existing and future -- and that this change in HCD methodology was the double-count the Embarcadero Institute reported in its study.

In fact, before SB-828 became law, HCD applied no overcrowding or cost-burdening adjustment at all, and our findings of a double-count are unrelated to this.

Fulton:

“Prior to SB-828 HCD applied these factors only to the new increment of housing called for in the RHNA. Post SB-828, HCD applies those factors to all housing, in an attempt to capture pent-up housing demand. Obviously, this change gooses the RHNA numbers. This is the supposed double counting.”

Separately, Stephens, in his article, admits that he didn’t fully understand the calculations. The misunderstandings in both Fulton’s and Stephens’ articles need to be clarified. 

Here’s what actually happened to create the double-count:

In 2015, during the preparation of HCD’s 2015 Statewide Housing Plan, the Dept. of Finance (DOF) convened a panel of experts, which recommended an adjustment to the DOF household projection model to compensate for the overcrowding and cost-burdening impacts of the Great Recession and affordability crisis. Those adjusted household numbers were then adjusted a second time for the same impacts by the HCD. These were dual actions of two departments in Sacramento who separately, and we understand unknowingly, added the same adjustments to the housing needs data. This is the double-count found by Embarcadero Institute.

• Prior to SB-828 becoming law, HCD did not make any overcrowding and cost-burdening adjustments. HCD adjustments for overcrowding and cost-burdening are new in the current cycle. They were not previously applied to housing estimates, future or existing.

• Post SB-828, HCD applied the overcrowding adjustment to all housing estimates, future and existing. They applied the cost-burdening adjustment only to future housing. In the case of the San Diego region, they did not apply a cost-burdening adjustment at all, even though the San Diego region is more cost-burdened than the Bay Area, according to the Comprehensive Housing Affordability Strategy data published by HUD.

• The double-count we identified is not related to whether HCD applied its adjustments to future or existing housing. The Department of Finance (DOF) adjusted household projections for overcrowding and cost-burdening before they passed their projections to HCD. Thus the factors were “baked-in” to the household projections from DOF. The subsequent adjustment for those same variables by the HCD resulted in a double-count.

2.  Why the RHNA Numbers Matter

Fulton and Stephens both argue that, regardless of the state’s erroneous double-count, pointing out this error is more obstructionist than helpful. Fulton argues that distinguishing the affordable housing needs from the total housing production challenge, is a distraction because:

“there’s not enough money in the world for California to solve the housing problem by government subsidies alone.”  Fulton argues that although “there’s no question market-rate housing production is not producing housing at all necessary price points, that doesn’t mean that we should give up entirely on the idea” that it eventually will.

Stephens argues that none of the state’s housing targets matter. Whether the housing need is for 1.17 million units (the estimate using the pre-SB 828 approach by HCD) or 3.5 million housing units (McKinsey & Company’s number), Stephens writes,

“the discrepancy doesn’t matter’” and the “debate is somewhere between harmless and pointless”  because:

a) we just need to know the number is really big; b) the amount of housing that is built will depend on capital, the availability of labor, the virus, and maybe Biden; and c) the RHNA housing targets are just zoning requirements and don’t require any housing be built. 

There is a Price to Be Paid: the RHNA Targets Aren't Just Suggestions

To an engineer, the words ”the numbers don’t matter” signal cognitive dissonance. Gauging the size of California's affordability crisis does matter, especially when the problem has been cited as California's most important. It particularly matters when the current state methodology produces an estimate that is more than twice that produced by previous state methodology (a methodology that was in use for two decades). And it matters because the state can enforce the targets through punitive action. There are two ways in which cities are now held accountable for the housing targets:

1. If cities don’t show they have rezoned for the housing mandated by the state RHNA targets, the state can withhold funding. Cities with non-compliant Housing Elements also risk being sued by the state, which can result in fines of up to $100,000 per month. Cities that remain out of compliance for six months can be fined as much as $600,000 per month. For some smaller cities, rezoning to make way for artificially inflated housing targets is particularly challenging because they have fewer options.

2. In addition, if a city fails to issue the number of permits by income level, as ordered by the state, housing developments at that income level are eligible for fast-tracked approval. In the past, because cities have easily exceeded their market-rate targets, only affordable housing projects would have been granted streamlined approval. Now, with exaggerated market-rate targets that have doubled, and in some cases, tripled under SB-828, suddenly market-rate housing could also qualify for special treatment. In a scenario where both market-rate and affordable housing have fast-tracked status, market-rate housing will be built preferentially because it is more profitable. So, yes the RHNA numbers matter.

Establishing a defensible and realistic methodology for the RHNA process is critical to ensuring the targets are seen as credible, and to ensure that there's a reliable measure against which to measure progress. Inaccuracies in HCD's methodology, caused in part by legislation, guarantee that cities will push back, as evidenced by litigation already underway. 

Why Cities Can’t Rely Solely on Market-Rate Housing Production 

To Fulton’s point, that we can’t entirely give up on the idea that market-rate housing production will eventually help solve the affordable housing crisis, we haven’t. Embarcadero Institute acknowledged the role of market-rate housing in the production of affordable housing. In fact, inclusionary zoning (where bonuses and incentives are granted to a market-rate development if it includes a percentage of affordable housing units) is responsible for the lion’s share of the affordable housing produced by cities. It’s practically the only tool cities have at their disposal. However, inclusionary zoning can only get cities so far and leaves them well short of their affordability targets (60% of all housing) as set by the state. Fulton argues that cities can’t rely on subsidies alone; Embarcadero Institute suggests cities aren’t relying on state subsidies at all. State subsidies for affordable housing dropped off a cliff in 2010 when the state shuttered local redevelopment agencies. Our argument is that cities can’t get anywhere near 60% affordability with virtually no subsidies from the state. 

The State’s Methodology Should be Sound

Stephens says he “got the sense that Embarcadero Institute thinks it’s pretty clever” in pointing out these calculation errors. We don’t. There’s nothing “clever” about close reading and basic arithmetic. The methodology used in the previous HCD model, in the new HCD 6th cycle model, and in the McKinsey model varies so greatly that we believe there is value in understanding the underlying assumptions and evaluating them for common sense and mathematical soundness.

Any state-adopted methodology should be defensible, agnostic to politics, and should not be set up to deliver an artificially high or low result. If these numbers are the basis for state housing policy and drive state budget allocations, that’s the least the tax-paying public can expect. 

Consider the McKinsey & Company analysis for example. I worked at McKinsey & Company for a number of years and know well that some great analysis has been produced by the firm, but their report on housing in California completely dismisses the work of the professional demographers at the DOF and their multifactorial population and household growth models that take into account net migration, births, deaths, household formation rates, and age cohorts. Instead, McKinsey & Company uses a simple back-of-the-envelope calculation,  multiplying New York State’s housing per capita by California’s population to determine the state’s housing need.  If McKinsey's approach is to be the basis for state housing policy then perhaps the demographers at the Dept. of Finance should pack up and go home.

3. Pressing Questions About Affordable Housing

Embarcadero Institute continues to think it is important to consider affordable housing and housing production as two separate issues, in part because the mechanisms for financing affordable housing are separate and distinct. On the supply side, affordable housing is financed through mechanisms like low-income housing credits, tax-exempt housing bonds, government-insured mortgages, down payment assistance programs, and direct funding. On the demand side, affordable housing is managed through waiting lists and supplemented with federal housing vouchers. The affordable housing market is far more constrained than the market-rate housing market and understanding that seems important to developing solutions to the housing crisis. 

This is particularly true as the two markets continue to diverge: the spread between the mean and median income (a rough measure of socio-economic inequality) continues to increase in California’s major planning regions. As it stands today in San Francisco, the average household income in the top 20% of earners is $345,000, while the average household income in the city’s bottom 20% of earners is $13,000. Combine this increasing income divide with the fact that RHNA cycle after RHNA cycle, the HCD sets affordable housing targets at 60% of the housing need (where “affordable” is extremely low-, very low-, low- and moderate-income housing). Yet cycle after cycle, only a fraction of that affordable housing is built. Data from HCD progress reports show that California is digging an increasingly bigger hole in the shortfall of affordable housing, yet most cities and counties are easily exceeding their market-rate housing targets from the state. 

The major reason cities aren’t approving enough affordable housing permits to meet their required targets is because they have no funding to subsidize affordable housing. Instead, cities primarily rely on the approach suggested by Fulton where affordable housing rides the coattails of market-rate housing. Developers are offered incentives like “density bonuses” that allow larger housing projects if the developer agrees to make 10% to 20% of the units affordable. It is not surprising then that those same percentages, 10% to 20%, mirror the amount of affordable housing approved by cities. 

Without state subsidies for affordable housing, it’s unreasonable to expect cities to achieve more. The real question is why Sacramento, having claimed housing is the state’s No. 1 priority, devotes less than 0.5% of the state budget to help cities fund the affordable housing quotas that it sets. We agree with Fulton’s assessment that California can’t solve the affordability problem by subsidy alone. We’re saying that when 60% of the housing a city has to build is affordable, it can't be done without subsidy.  

4. Deeper Dive into the Dept. of Finance Methodology

Dept. of Finance (DOF) explains the methodology behind its adjustment in the “read me” section of its Projection Tables P-4 (see excerpt below). In the past, DOF’s model was based on 20-year household headship trends. However, in 2015, concerned that the impacts of the Great Recession would disproportionately affect the 20-year trend data, the state convened an expert panel that decided to make a recession adjustment to the model. The adjustment was meant to correct for overcrowding and cost-burdening as a result of the recessed economy -- effects such as the doubling up of households, and young adults delaying forming households on their own. The DOF’s new approach results in higher household forecasts and essentially benchmarks household formation to conditions that, as the DOF describes it, reflect socio-cultural norms of home-ownership and household size.

Dr. Walter Schwarm, Chief Economist at DOF, explains in his “read me” notes: 

“Age- and race/ethnicity-specific headship rates were calculated based on 1990, 2000, and 2010 census information. Trends in headship rates over time by age and county are modeled and used to project future headship rates. In addition to trends, there is an explicit return to an average of 2000 and 2010 headship rates built into the model. This average of rates was the result of deliberations made by the expert panel during the previous HCD Statewide Housing Plan in 2015. The argument was that the Great Recession and the affordability crisis which impacted recent trends in headship should not be allowed to solely dominate the projection, rather some return to underlying socio-cultural norms of homeownership/fewer roommates is a beneficial assumption that reflects the fact that those conditions were temporary.”

5. Additional Dept. of Housing and Community Development Anomalies 

• The HCD overcrowding adjustment is inaccurate, in assuming that 1 overcrowded home equals the need for 1 additional housing unit. Overcrowding is defined as more residents than the number of rooms in a house (excluding bathrooms). In many cases (based on American Community Survey data), an overcrowded household is overcrowded by 1 or 2 people. A more logical approach would be to determine the number of additional people requiring housing, and from that, determine the number of households needed. 

• The new state methodology varies from region to region, e.g. there was no cost-burdening adjustment to San Diego even though its cost-burdening statistics, according to numbers published by the HUD, were higher than the Bay Area. National benchmarks were used for the six counties of Southern California, the Sacramento region, and the San Diego region, whereas the Bay Area was benchmarked against the seven largest Combined Statistical Areas (CSAs), generating very different outcomes. Should large urban areas be benchmarked against national averages that include rural areas?

• There is a potential double-counting between overcrowding and cost-burdening, as the households that are overcrowded are largely low-income households, which are also the same households that are cost-burdened. 

• Overcrowding adjustments were applied to future housing as well as existing housing. Cost-burdening was only applied to future housing. Both assume the conditions continue to exist in some unknown future, despite the fact the targets are meant to address these conditions.

6. Additional Corrections from Stephens' article

1. “Conventional economist approach”: Stephens didn’t understand what Embarcadero Institute meant by that, and he wasn’t alone. We used the term to denote HCD’s previous method, used for two decades before they changed their formula based on SB-828 requirements. HCD’s previous method is also described by Harvard’s Joint Center for Housing Studies: 

   “Housing need =  projected households + natural vacancy + replacement - existing housing units.”

   We refer to it as “conventional” in our report because it is used and accepted by so many academic institutions and demographers.

2. Housing shortfall versus housing need: Neither 1.17M units (the estimated housing need produced by HCD’s previous long-term model) nor 3.5M units (McKinsey & Company’s estimated housing need) are measurements of housing shortfall, as Stephens assumes. Both numbers include the estimated existing housing shortfall PLUS future housing needs.

1. Job-to-housing ratio: Stephens wrote that the ”Embarcadero report helpfully cites a homes-to-jobs ratio of 1.5 … as if dependent or unemployed people miraculously don’t need homes.”

   This is incorrect. In fact, we cited a ”jobs-to-housing” ratio, not a “homes-to-jobs” ratio. Jobs-to-housing is a standard planning metric. The American Planning Association says that a healthy range in jobs-to-housing is 1.3 to 1.7. The Building Industry Association (BIA) says it’s 1.5, Association of Bay Area Governments (ABAG) says it’s 1.41.  We used the middle of the range. None of these bodies dismissed the obvious need to account for the housing of dependents and unemployed people. The denominator, housing units, includes all types of households. These ratios and the healthy ranges are debated in peer-reviewed academic journals. Embarcadero Institute did not invent the idea.