This content was first published by the ADP Research Institute on November 14, 2024

More than a million people in the United States earn $500,000 or more, and they might be closer to home than you think.

High salaries consistently capture public attention, putting multi-million-dollar compensation in the spotlight. But while these exceptionally high earners might be rare, high-paying jobs aren’t. An analysis of ADP payroll data reveals that about 1 in 127 jobs in the United States pays more than $500,000 a year, and the reality of these high wages is closer to home than you might think.

It can be difficult to get a clear picture of just how prevalent massive paychecks are. Government data sets typically obscure the highest incomes by capping the level of wages reported. The Census Bureau’s 2022 American Community Survey, for example, set income caps at various levels less than $1 million. In half of U.S. states, the caps that year were even lower—less than $500,000.1 This practice protects high earners from being identified publicly, but it also limits our understanding of exceptionally high wages.

Anonymized payroll data from ADP allows us to get a better view of exceptionally high wages. Because it’s based on administrative records, ADP data reflects a much larger share of the population than Census survey samples. And while ADP data doesn’t represent a random sample of earners, we have adjusted it to reflect the full population and provide a much more detailed view of top earners than publicly available government data.

What we found

What’s striking is how many people in the United States earn exceptionally high salaries. While just 0.79 percent of jobs in the country paid more than $500,000 per year, that’s well more than 1 million positions.2 This isn’t just a small handful of executives—it’s a substantial number of professionals found in every major metro across the country.

Certain regions stand out for the sheer volume of workers pulling in substantial incomes. A ranking of large U.S. metros by the share of residents working at jobs that paid $500,000 or more has Austin and New York near the top. But when it comes to high-paying jobs, one region is in a league of its own.

Silicon Valley exceptionalism

The San Francisco Bay Area, home to Silicon Valley’s tech giants, vastly outranked every other metro in terms of high-paying jobs, including those paying more than $1 million or $2 million annually. About 1 in 48 Bay Area jobs paying more than $500,000 a year, nearly double the share in the second-highest region, Austin.

What drives this exceptional concentration of high earnings? Unlike other highly paid professions—such as doctors or lawyers, whose income is limited by the number of patients or clients they can serve—productivity in the tech industry, particularly at large employers, faces fewer constraints. A single star engineer or executive has the power to dramatically shift a company’s trajectory, and tech giants with enormous revenue at stake often reward these select individuals with extraordinary compensation.

In that sense, it’s no coincidence that tech-heavy Austin is second to Silicon Valley, followed by New York, which combines a strong tech presence with its role as the world’s foremost financial center. The Bay Area’s considerable lead likely reflects not just the dominance of tech in its economy and workforce, but also its position as the nucleus for the industry’s top talent and the corporate giants that rely most heavily on their expertise.

Compensation for star tech talent isn’t the only factor at play. Local housing costs play a critical role in determining who can afford to live where, and nowhere is that more a factor than in the Bay Area. Decades of skyrocketing housing prices have squeezed out low- and middle-income residents and discouraged potential newcomers. As a result, high earners account for a greater share of the population than they would otherwise.

The rise of remote work has given people greater freedom to live in one place while working for an employer elsewhere, hollowing out the middle of the Bay Area’s wage distribution and attracting high earners to other desirable locations. The relative abundance of people making $1 million or more in Honolulu, Las Vegas, parts of Florida, and affluent rural areas likely reflects a mix of natural amenities, lifestyle appeal, or favorable tax policies.

Despite tech’s dominance in shaping high-wage hubs, very high-paying positions are found in substantial numbers across every major U.S. metro. Whether driven by finance in New York, healthcare in Boston, or energy in Houston, high earners aren’t confined to one industry or region. Though tech is at the forefront, very high salaries are more prevalent than one might realize—perhaps just down the street, or down the line.

Methodology

We used a sample consisting of hundreds of millions of active and anonymized ADP payroll records spanning the period from July 1, 2023, to June 30, 2024. To address uneven pay over the calendar, annual pay amounts reported correspond to 12 times the average monthly pay over the prior 12 months, or over the duration to date of the current employment spell if less than 12 months. An entry from July 2023 reflects pay over the 12-month period ending that month, for example. Pay amounts reflect gross pay as reported by employers, which refers to pre-tax pay including bonuses, commissions and equity-based pay such as stock grants and options, as well as subsequent employee deductions for benefits. Records were adjusted for pay periodicity, for example monthly vs bi-weekly, and partial months of employment were omitted. Observations were re-weighted to be more nationally representative using information on local employment by industry and firm size from the BLS Quarterly Census of Employment and Wages. Metropolitan areas reflect workers’ location of residence. They are limited to metros with more than 1 million residents. Metro area definitions correspond to Consolidated Statistical Areas where available and Core-Based Statistical Areas otherwise.

Notes:

1. The highest reported income cap in the 2022 American Community Survey was for Washington, D.C., at $791,000 a year. The lowest was for Mississippi, at $363,000 a year. The ACS currently sets the cap at the 99.5th percentile of each state’s pre-tax wage and salary income distribution.

2. Per the Quarterly Census of Employment and Wages, average annual employment in the United States during 2023 was 153,140,899, or 131,289,681 if only the private sector is included. For both numbers, 0.79 percent is greater than 1 million.

This content was first published by the ADP Research Institute on May 22, 2024

Long-distance work is splitting geographic cost savings between workers and employers and improving the match between them.

Long-distance work didn’t begin with the coronavirus lockdown; people were collaborating between offices well before then. But four years after the pandemic’s normalization-by-necessity of remote work, many long-distance workers never went back, and their ranks are growing.[1]

So, does remote work pay? And if it does, for whom? Employees who go remote can raise their standard of living overnight by moving to a city with less expensive housing and an overall lower cost of living. But our research found that employers can share in the geographic cost savings, too. Perhaps most important, the widespread acceptance of long-distance labor has improved the odds that workers and jobs will be well-matched, which stands to help the labor market as a whole.

In earlier studies, we examined cross-metropolitan work—in which the worker and the manager live in different metro areas—to document the unprecedented scale and scope of long-distance work and its geographic implications for cities.[2] In this study, we examined payroll and HR data of some 1.3 million workers each month to see how the rise of long-distance work has influenced compensation.[3]

We found that job-stayers who stayed with their team but relocated to less-expensive communities had slower average wage growth than their on-site teammates. That discount suggests that employees are sharing the geographic cost savings of relocation with their employers.

In contrast, people newly hired into long-distance positions won bigger average pay gains than similar job-switchers who took local positions. In some cases, that premium reflects the transition from local work to long-distance work for employers located in higher-paying markets. But we found bigger average pay gains for long-distance workers hired to companies in lower paying markets as well, which suggests something different altogether: Long-distance positions might be helping both workers and employers find a better fit with each other.

The big picture is that the normalization of remote work has created opportunities for employers and workers alike and has the potential to make the labor market more efficient by better matching the right person to the right job.

Sharing the benefits of long-distance work

People can transition into long-distance jobs in multiple ways. The most straightforward scenario is one in which a person relocates without giving up their current job.[4] When a worker moves to a community with cheaper housing and a lower cost of living, do they maintain their pay level and reap the full geographic cost savings, or does their pay take a hit, giving their employer part of the savings?

We tracked the gross wages of job-stayers who transitioned from local to long-distance work by relocating to communities with cheaper housing. We compared their earnings in the months before and after the move.[5,6,7] We did the same for teammates who didn’t relocate.[8] We statistically controlled for factors such as personal attributes, job function, and location to obtain an approximate yet meaningful apples-to-apples comparison of average wage growth.[9]

In the four years preceding the pandemic, from 2016 to 2019, average wage growth for job-stayers who relocated to affordability was 0.3 percentage points below that of their teammates, a difference statistically indistinguishable from zero. The difference remained indistinguishable from zero for 2020-2021.

But for 2022-2023, the wage growth deficit increased to 2.8 percentage points.[10] Given that the median pay gain for job-stayers was 7.3 percent in 2022 and 5.2 percent in 2023, this 2.8 percentage point deficit is considerable.[11,12]

While some employers might have recalibrated workers’ pay in proportion to cost-of-living differences, most did not, because cost-of-living differences are generally far more substantial than the wage growth deficit we found.[13] That means employers, at least on average, are splitting the geographic savings that flow from relocation with existing workers, with the lion’s share going to the latter.

Finding the right fit

Reducing an employee’s pay can be difficult; setting the wages of new employees less so. With that in mind, employers can better capitalize on geographic cost savings by hiring new long-distance workers for far less than they pay their existing workers.

As with our job-stayers exercise, we used ADP data to track the gross wages of job-switchers who transitioned into long-distance work and compared their wages before and after the transition.[14] Unlike job-stayers, who generally transition into long-distance work by relocating, job-switchers can transition into long-distance work with or without relocating.

We compared the wage growth of job-switchers into long-distance work with that of job-switchers who took new positions locally with the same employers.[15]

Here, too, new hires can vary in terms of personal attributes such as age and income level, job function, and location. The tendency to transition into long-distance work can correlate with those factors, so we controlled for them to obtain a meaningful apples-to-apples comparison.[16]

Job-switchers hired into long-distance roles saw substantially greater wage growth than similar job-switchers who took local positions with the same employers.

Since the pandemic, this gap has widened substantially. From 2016 to 2019, wage growth for people hired into long-distance work was more than 8 percentage points higher than for similar people who took on new jobs locally with the same employers.[17] By 2022-2023, the difference had risen to nearly 16 percentage points.

Why might the match between workers and jobs matter for pay?[18] Increasing the number and variety of job openings and candidates available—a thickening of the labor market— increases the potential for workers and employers to find better matches than they could otherwise.[19]

The explosion of long-distance work has significantly expanded the pool of jobs that applicants can consider and has enabled people to find better-matching jobs than they might otherwise. It also has helped employers find better candidates.[20] A better match ultimately creates value, which can manifest with improved productivity, greater job satisfaction, lower turnover, higher employer profits, and greater worker pay.[21]

And one worker-job match may be better than another because the applicant lives in a less expensive area and is willing to accept less pay. The post-pandemic jump in the wage growth differential of job-switchers into long-distance work can be attributed to geographic cost savings as well as better matching in a thicker market.

How might we separate the two? If the alternative to hire a long-distance worker is to hire locally, near the manager, then a long-distance worker-job match has the potential for geographic cost savings only when workers live in less-expensive places than their managers, not the other way around. If we examine job-switchers who take on long-distance roles reporting to managers in both more- and less-expensive places, any wage growth differential of the latter can be attributed only to better matching.[22] Figure 3 repeats the job-switcher exercise for those two groups.[23]

The wage growth differential for job-switchers taking long-distance work increased for both groups but was bigger for people reporting to managers in more-expensive areas. This result is consistent with that greater wage growth differential reflecting both geographic cost savings and better matching.[24]

Crucially, however, we also see an increase in the wage growth differential for remote job-switchers who report to managers in less-expensive places, where there are no geographic cost savings to be had. This higher relative wage growth suggests that long-distance work began producing better worker-job matches after the pandemic. The thickening of the market for long-distance work—the greater availability of non-local candidates and job openings due to the normalization of remote work—strengthened the match-quality advantage of long-distance work.

Looking ahead

Improved worker-job matching can benefit the economy and society because people who are in the right jobs tend to be more productive and satisfied.[25]

Long-distance work has its problems, including isolation and the challenges of remote managing, but its increasing prevalence suggests that the benefits outweigh the frictions.[26,27]

As long-distance workers multiply and gain experience, new career trajectories might emerge. People could transition between in-person and long-distance careers, for example, the way lawyers move between partner- and non-partner career tracks, reflecting a trade-off between higher pay and better work-life balance.

And while both substance and presentation will continue to matter, long-distance work might lend itself to showcasing one’s work more prominently than one’s social skills, perhaps taking the market closer to meritocracy.

But as one of our previous studies noted, lack of proximity may come at a cost to society. The tremendous growth of long-distance work has turbocharged the process of domestic offshoring, which concentrates value-intensive jobs in the nation’s most expensive cities while sending rank-and-file ones to more affordable parts of the country. Inasmuch as opportunity depends on geographic proximity and interaction in-person, long-distance work that fuels domestic offshoring runs the risk of separating most Americans from opportunity by a barrier of distance. The many advantages of long-distance work may or may not be enough to surpass that barrier.

Notes:

1. We use the terms long-distance work and cross-metropolitan work synonymously to describe the relationship between a worker and a manger who live in different metro areas. Those terms are distinct from remote work, which refers to work that is regularly performed off-site, regardless of which metro the worker or their manager reside in. For more details, see the first study in this series.

2. The degree of employment interconnectivity between U.S. cities has exploded since 2020, and that has turbocharged the process of domestic offshoring, whereby value-intensive jobs are increasingly concentrated in the nation’s most expensive cities, while rank-and-file positions locate in more affordable parts of the country.

3. This study and earlier studies in the series employ a sample of ADP payroll and HR records corresponding to subset of employers serviced by ADP with at least 1,000 employees in the United States. Within that employer group, the sample was further limited to individuals who are members of teams with a reliable set of reporting structure fields by limiting the data to teams in the United States in which no more than 10 employees report to the same manager. Although the number varies over time, approximately 1.33 million employees at 3,500 businesses, on average, met these criteria each month from January 2016 to June 2023.

4. A worker can also transition into long-distance status passively if their local manager relocates to another metro area, or actively by taking on a new long-distance job, i.e. one in which they reporting to a manager living in another metro area, with or without relocating themselves.

5. Following the logic of event study models, we determined workers’ pre- and post-transition metro of residence, employer, and team identity (as reflected by the identity of the manager), by considering a six-month window before and after each month of observation (the current month of observation was included in the after window). Only records with fully observed before and after windows (i.e. with six consecutive monthly observations, containing all necessary information), and that were also internally consistent with respect to the metro, employer, and team identity were kept in the sample. Cases in which a manager’s location changed during the before or after window or between the two were omitted as well to better ensure continuity of the worker’s role. Finally, records were omitted when they corresponded to a job spell with average monthly wages below a $20,000 annual income, when they overlapped with another observed employment spell for the same worker having higher average earnings (i.e. when a record did not belong to a worker’s primary job), or when they reflected the first or last month of an employment spell, when earnings are likely to reflect only a partial month. Wage growth was obtained as the percent difference between the average gross wages in the last five months of the after window and the first five months of the before window. The current and preceding months were omitted from the wage growth calculation to avoid assigning pre- and post-transition pay incorrectly given uncertainty around the exact timing of the relocation and the potential wage update within those two months. To eliminate outliers, wage growth for job-stayers was truncated below halving and above doubling.

6. Relocation was determined based on metro areas defined broadly. Specifically, metropolitan area definitions used for determining relocation and long-distance status (i.e. for comparing workers’ and managers’ metro of residence) refer to Consolidated Statistical Areas (CSAs) where those are applicable, and to Core-Based Statistical Areas (CBSAs) elsewhere. By using the broadest available definitions of metropolitan areas, we minimize situations in which workers separated by arbitrary lines within broader metros, e.g. between the San Francisco and San Jose metro areas. The downside of this approach is that the largest metro areas internalize employment relationships that are cross-metro in kind. In contrast, relocation up and down the housing price gradient was determined based on workers’ narrowly defined metro area, i.e. CBSAs. That is a pragmatic choice, because housing price indices are readily available for CBSAs but not CSAs. It also allows us to better capture the actual local housing cost levels faced by individuals. Housing price levels associated with each CBSA were obtained from the public-facing Zillow Home Value Index. To avoid muddying inter-metropolitan housing price comparisons with housing price growth during the years observed, we assigned each metro area its housing prices from February 2020—on the eve of the pandemic—regardless of the timing of the associated record in the data. Relocations were designated as being up the housing price gradient when the origin to destination housing price difference was greater than 10 percent, and as down the housing price gradient when it was below negative 10 percent.

7. We measured near-immediate wage growth around the transition. It is possible that workers transitioning into long-distance status would experience a different trajectory of wage growth than they would otherwise over time. In that case, the changes in their wage growth owing to the transition could compound gradually over time, causing them to deviate from the near-immediate wage growth that we observe.

8. The estimates were performed via linear regression with team fixed effects. For details, see regression appendix.

9. We know from an earlier study, for example, that the age of workers flowing into long-distance status changed markedly during the pandemic. Inasmuch as wage growth tends to steepen or flatten at certain points of the lifetime career trajectory, a changing mix of ages among those transitioning into long-distance work could influence their observed wage growth relative to those staying local. For details of the controls used, see regression appendix.

10. In contrast, job-stayers transitioning into long-distance work by relocating up the housing price gradient, i.e. to more expensive places, experienced average wage growth around the transition that was about 2 percentage points below that of their teammates in 2016-2019. That difference turned positive in 2020-2021, then reverted to its 2016-2019 level in 2022-2023. It is not obvious how to interpret that result. The key probably is to ask what kind of selection determines who moved up the housing price gradient within jobs in each period. Internal transitions up the housing price gradient pre-pandemic were rather uncommon and perhaps included an over-represented contingent of relatively junior employees who were more tolerant to expensive cities prior to having children, and whose wage growth may have been lower than that of their more senior peers. During 2020-2021—but much less so in 2022-2023—such moves probably included a very different pool of people, as the typical motivation suddenly became living near family again, including for workers whose families are located in expensive places.

11. The median pay changes for job-stayers correspond to year-over-year changes as of January 2023 and January 2024, drawn from the ADP Research Institute’s Pay Insights report as of February 28, 2024.

12. As noted in footnote 5, we are measuring the near-immediate wage growth deficit around the transition. If that wage growth deficit were to persist over time and compound, it could become even more meaningful.

13. Housing costs—a proxy for the cost of living—are highly dispersed across metros. For example, the interquartile range across CBSAs of the February 2020 Zillow Home Value Index is a 2.4-fold multiple.

14. This exercise followed the same lines as the previous one, as detailed in footnote 5, with some differences. Here we used a three-month window before and after each observation to determine the pre- and post-transition metro of residence and employer identity for workers and their distinct managers. We also allowed up to a six-month gap between the end of the pre-transition job and the beginning of the post-transition one, using the first month of the post-transition spell to time the transition. Overlapping jobs held by the same individuals were more fully purged, with any job spell observed to overlap with another removed in its entirety to reduce instances of part-time roles populating transitions. To eliminate outliers, wage growth for job-stayers was truncated below two-thirds reductions and above tripling (reflecting the wider wage swings of job-switchers relative to job-stayers).

15. Estimates were performed via linear regression with post-transition employer fixed effects.

16. The controls used here are the same as those used in the regression informing Figure 1, with one exception. The previous regression controlled for job-stayer location only prior to the transition to avoid inadvertently controlling for the potential for geographic cost savings. The current regression controls for the pairing of job-switchers’ pre- and-post transition locations, because in this case the employer’s potential for geographic cost savings depends largely on the housing price gradient between the manager’s location and the worker’s location. For job-stayers transitioning from local to long-distance work, the origin-destination housing price gradient is generally identical to the manager-worker housing price gradient, but for job-switchers into long-distance work, the two are often different. Here, we control for the worker’s origin-destination housing price gradient which can differ across workers joining the same employer and could potentially correlate with post-transition long-distance status, but we don’t control for the manager-worker housing price gradient, which is the main determinant of the employer’s geographic cost savings. For more details of the controls used, see regression appendix.

17. The wage growth differential for job-switchers transitioning into long-distance versus local work could be attributed to better matching in a thicker market even before the pandemic. The normalization of remote work during the pandemic changed that not qualitatively, but quantitatively: By vastly increasing the number of long-distance job openings and candidates, it thickened the market, increasing match quality and raising the premium for long-distance work. That said, the pre-pandemic result could be attributed, at least in part, to omitted variable bias, as the controls included in the regression cannot ultimately hold all else fixed.

18. It is beyond the scope of this study to identify and disentangle with certainty all the causes of the post-pandemic jump in the excess wage growth of job-switchers into long-distance work, however we can make some headway. First, the jump could stem from changes in the mix of long-distance employers. That is why we compare job-switchers’ transitioning into long-distance work with those switching into local work for the same employers. Second, the jump could stem from changes in the mix of long-distance workers. That is why we control for worker attributes, including their locations before and after the job transition (See footnote 16, as well as the regression appendix). Finally, the jump could stem from changes that the reflect the pairing of jobs and workers, i.e. the match between them. That is what we are concerned with.

19. A prerequisite is that workers and employers are able to search for each other effectively, i.e., that they should be able to sift through the vast pool of potential partners and find those that offer the best match, or at least a better one than they might find otherwise. Search can be conducted effectively today only to the extent that it is well-mediated by the internet, for example through online job and hiring search engines. In some ways, that is true for localized search as well, though it seems likely that there is a benefit to greater local familiarity that might be lost in long-distance search.

20. Strictly speaking, the normalization of remote work since the pandemic increased the number and variety of job openings and candidates, but it might have limited local options for workers and employers because of competition. That limitation of options is likely to be the most noticeable for workers in the most expensive places, who might have fewer local job options as local employers opted to hire remotely, and for employers in places with the lowest wages, who might have lost local candidates to competition from remote employers. That said, because all local labor markets are small compared to the national (or global) labor market, the increase in remote options seems likely to dominate any reduction in local options nearly everywhere. All participants experienced a thickening.

21. Better matching is not confined to the kinds of skills and job requirements that might appear on a resume or job description. It also includes myriad other criteria that matter for the people involved, such as their schedules, habits, expertise, and more.

22. Note that in the case of job-stayers—workers relocating while keeping their existing jobs—the geographic aspect of the match changed while all other aspects remained largely unchanged.

23. See footnote 4 for details on the assignment of housing price levels to CBSAs. Worker-manager relationships were designated as being up the housing price gradient when the housing price difference between the manager’s CBSA and the worker’s was greater than ten percent, and as down the housing price gradient when it was below negative ten percent.

24. Specifically, the excess wage growth of job-switchers into long-distance work up the housing price gradient reflects the worker’s portion of the geographic cost savings and the improved matching. Presumably, the employer benefits from both, too.

25. In fact, even for workers and employers who do not engage in long-distance work directly, the age of remote work could shape the set of outside options, affecting them indirectly. For example, if local workers have a new option of earning more from distant employers, local employers trying to retain a sufficient workforce might need to raise wages for employees who continue working locally. This can hold even in lines of work which don’t lend themselves to being performed remotely as long as worker options include the ability to transition to other lines of work that do. Similarly, if employers have a new option of hiring distant workers at lower cost, that might exert downward pressure on local workers’ wages. The potential indirect effects of the new bargain on workers and employers not engaging directly in long-distance work are closely related to the factor-price equalization theorem in the Heckscher-Ohlin model of comparative advantage in trade. And while such factor-price equalization with respect to wages in different places is unlikely to occur in isolation—the world is more complicated, as illustrated by the evidence of domestic offshoring—it nevertheless is likely to be an important element in what unfolds.

26. Sensitivity to those frictions can vary significantly across different types of work. As we argued in a previous study, the degree of sensitivity to those frictions is likely a key element in determining the pattern of domestic offshoring, i.e. the geographic sorting of work between the groups of U.S. cities we dubbed leadership hubs and workforce nodes.

27. Long-distance work offers individuals at least two non-pecuniary benefits: Optionality and opportunity. Optionality is improved because a broad swath of jobs are more geographically footloose and job choices are less dependent on where one lives. With limitations, a Wall Street career can now be had in the Midwest, permitting workers to separate their careers from the needs, wants, and circumstances that determine where they live. Opportunity is another benefit. A thicker market means workers searching for jobs will have more gold nuggets to find, so to speak, if they’re able to tackle a more complicated search process. Those who engage the long-distance job market simply by applying to the same handful of superstar firms as everyone else likely will be wasting their time. New strategies and wisdom will emerge around job search, which is likely to be a higher ROI activity than it used to be. Employers, too, will need to adapt recruitment in the long-distance job market. Familiarity with the nation’s best universities will not suffice; employers will need to familiarize themselves with sources of talent that more local and less known.

Regression Appendix

This paper presents a new approach to measuring housing affordability—one that seeks to provide a better indicator of what counties in California are affordable, and for whom. The authors propose a new inclusive measure that considers whether a location is affordable not to those who live there, but rather to those who might potentially live there. The authors begin by explaining the problems with the way current affordability measures are constructed, and detail how we address those problems in our alternative inclusive affordability measure.

The paper is paired with Affordability for Whom?—a web-based interactive tool that enables detailed examination of how this measure impacts affordability at the county level.

This content was first published by the ADP Research Institute on March 7, 2024

In 2020, the Covid-19 pandemic ushered in an unprecedented new era of remote work. Despite some institutional backtracking as health concerns have abated, remote work remains far more prevalent and acceptable today than it once was.[1]

That’s given employers access to a broader labor market and new options for configuring their talent pool—and its price tag. That is true whether the work is performed from a couch or from a satellite office.

Remote work eases access to new labor markets because it means the satellite office is no longer a prerequisite for employing workers in faraway places.[2] It also creates opportunities for workers to take their existing jobs elsewhere or take on work that once would have meant relocation.

In a prelude to this study, we used ADP data to show that the prevalence of long distance or cross-metro work—a close cousin of remote work—has skyrocketed since early 2020. In this study, we turn our attention to the proliferation of new employment connections between America’s cities. We look at the nature of those connections through the lens of company reporting structures, separately examining the location of workers and their managers.

One might expect remote work to level the playing field between places, but, perhaps counterintuitively, the mitigation of distance as a barrier to employment has accentuated some existing differences.

Working out of one metro and into another

To study employment connections between places, we need to know where workers are and where jobs are. Unfortunately, the location of remote jobs isn’t always well defined. We use ADP data on the reporting structures of large employers to pin each employment relationship to the worker’s metro of residence on one end and their manager’s on the other.[3,4] When the worker lives in a different metro than their manager, we call this long-distance work or, more analytically, cross-metro work out of the worker’s metro and into the manager’s. Those out- and in-bound relationships form the basis for this study.

We refer to metro areas that have more workers reporting into them than reporting out to managers elsewhere as leadership hubs, and to those with fewer in-bound than out-bound workers as workforce nodes. Each metro area’s place on the spectrum can be captured by a single number, its net leadership ratio, which divides the number of in-bound workers by the number of out-bound ones. A ratio greater than 1 indicates that a metro specializes in managerial tasks. A ratio of less than 1 indicates specialization in individual or frontline work.

Larger cities have long been net leadership hubs

Figure 1 shows the net leadership ratios of U.S. metros with populations of 250,000 or more.[5,6]

Large cities tend to have higher leadership ratios than small ones, as highlighted by the fitted lines which capture the average relationship between net leadership ratios and metro size.[7] That’s not surprising, and it’s consistent with the intuition that larger cities play a more central role in the economy. The relationship is clearest when metros of all sizes are pooled together, but it holds when the pool of metros is limited to those with populations greater than 1 million.[8,9]

Remote work has raised the leadership ratios of the most expensive cities

The data so far offers a snapshot of leadership ratios by metro as of January 2020, just before the pandemic upended life in the United States. The important question, however, is what the proliferation of remote work has changed since then.

In the years immediately before the pandemic, net leadership ratios didn’t change, on average. Changes that did occur didn’t vary systematically with city size or housing prices.

Figures 2a and 2b show the change in net leadership ratios for metros with at least 1 million residents from January 2017 to January 2020.[10] Figure 2a plots those changes against population, while Figure 2b plots them against typical metro home values.[11] The fitted lines capture the average relationship with population and home values. In both cases they’re close to zero and essentially flat.

Things changed with the rise of remote work. Figures 3a and 3b repeat the same exercise for January 2020 to January 2023. The relationship between the change in metro leadership ratios and population remained flat during the later period, but the relationship with home values turned significantly positive.

That means more expensive cities have, on average, become increasingly more specialized in managerial tasks since the onset of the pandemic, while more affordable ones have become increasingly more specialized in individual contributor and frontline work.

How much so? Suppose home values in one city are about twice as high as those in another. While leadership ratios for both cities would have fared similarly before the pandemic, the more expensive city’s ratio would have increased by about 5 percent more in the three-year period since. Leadership hubs have ratios of around 1.2 (Fig. 1), which means a 5 percent increase to 1.26 over three years is rapid and meaningful, especially if it continues to compound over time.

Domestic offshoring, turbocharged

The shift to prevalent remote work has rendered expensive cities increasingly management-heavy, while rank-and-file workers locate elsewhere.

That suggests that the rising leadership ratios of America’s most expensive metros are an invigorated extension of a longstanding pattern of domestic offshoring: The increasing ease with which workers can collaborate from distant places has shifted low-paying work to more affordable places and concentrated value-intensive jobs in expensive ones. That sorting—which began long before the pandemic—has been enabled by technology and propelled by the yawning gap in housing costs between America’s priciest cities and its more affordable ones.

What is keeping leadership concentrated in the expensive cities? Those cities have one big advantage: They enable decision-makers to interact with each other in ways that benefit from established in-person relationships precisely because those leaders are concentrated there. Headquarters, elite universities, research facilities, think tanks, clients, regulators, competitors, and the media all are disproportionately located in leadership hubs. A manager who is not ever-present in the right circles in those hubs will be challenged to develop the personal rapport and familiarity upon which influential decisions depend. Video conferencing and even frequent visits to a leadership hub are a poor substitute, especially when the stakes are at their highest.

For individual and frontline work, the stakes tend to be lower, making those jobs easier to perform remotely. Employers, therefore, have an incentive to locate those workers in places that are more affordable. Companies weigh the wage advantage of locating jobs in more affordable places against the frictions introduced by long-distance work, frictions that tend to be substantially greater for more influential roles.

By mitigating distance as a barrier to employment—much more so for rank-and-file workers than for managers—prevalent remote work is turbocharging the process of domestic offshoring from expensive leadership hubs to more affordable workforce nodes. In this sense, remote work is accentuating existing differences between metros.

Interconnectivity between cities is exploding

In addition to its role as a catalyst of domestic offshoring, the proliferation of remote work is also boosting employment interconnectivity between U.S. cities.

That means that the share of employment relationships that cross metro lines is exploding. In almost every city, a growing share of workers are reporting to managers located elsewhere, and managers in almost every city are responsible for workers who are located elsewhere in greater numbers.

Despite the growing leadership ratios in the most expensive cities, these urban centers also are seeing growth in the share of both in- and out-bound work, as are nearly all other cities.

How do we measure employment interconnectivity between cities? “Out-bound workers” describe people who are situated in one place and report to a manager who is situated somewhere else. “In-bound workers” are measured from the point of view of managers, a label that describes direct reports who live outside the manager’s metro. To measure a metro’s interconnectivity, we look at the share of a metro’s resident workers who are out-bound, and at the in-bound share among those reporting to managers in the metro.

Figure 4 shows the distribution across metros of the growth in each metro’s share of out- and in-bound workers, before and after the onset of the pandemic. The left panel indicates that, for the median metro, the out- and in-bound share of workers grew by 2.4 percent and 4.0 percent, respectively, from January 2017 to January 2020.

From January 2020 to January 2023, the ranks of remote workers grew by 31.3 percent and 35.0 percent—about 10 times faster. While the distribution included some metros whose share of out- and in-bound workers shrank before the pandemic, at least 90 percent of metros saw those shares increase after the pandemic.

In other words, before the pandemic, long-distance work was growing gradually, but since then it has been growing far more rapidly in almost every city. Even in expensive places like San Francisco and New York, a sizable share of workers reports to managers elsewhere.

The unprecedented growth of interconnectivity between cities is more than a statistic. It corresponds to new norms in the organization of work. There is now much less need to concentrate members of a team near the same office, and online communication increasingly is default way that work is done, even in on-site settings.

A new type of compensation strategy

The implications do not end with how meetings are managed.

The new world of distributed work means that employers in many instances no longer need to tie a job to a single place. A labor pool unbound by location offers the potential to lower labor costs while improving the fit between employers and candidates.

But even that’s just part of the story. When an employer ties a job to a place, it also determines the experience that job can offer employees. Will the job require workers to live far from their families, make new friends, and adapt to a different climate or culture?

In this new age of location-agnostic work, employers can grant employees the flexibility to choose the location that best suits their needs, wants, and circumstances. It’s not just about hiring in cheaper places, but about offering workers the optionality that they covet. What do employers get in return? The potential to reduce labor costs even in expensive places, while raising their appeal to talent.

For more on this thread, stay tuned for the next installment in this series.

Notes:

1. As of mid-2023, 28.1 percent of workdays across all lines of work (including those impossible to perform remotely) were worked at home. That figure was down from a peak of 61.5 percent at the onset of the COVID-19 pandemic, but up from a mere 7.2 percent immediately before the pandemic. See Figure 1 of Barrero, José María, Nicholas Bloom, and Steven J. Davis. “The evolution of work from home.” Journal of Economic Perspectives 37.4 (2023): 23-49.

2. While the technology that enables remote work is recent and necessary for it, it was available before the pandemic. Rather, it was the coordinated mass shift into prevalent remote work and the consequent social acceptability of it that were the most consequential impacts of the pandemic in this context.

3. The sample used consists of a subset of employers in ADP data with at least 1,000 employees in the U.S. Within those employers, the sample was further limited to members of teams with a reliable set of reporting structure fields by limiting the data to teams in the U.S. in which no more than 10 employees report to the same manager. Although the number varies over time, there are approximately 1.33 million employees that meet these criteria each month on average from January 2016 to June 2023, employed by an average of 3,500 firms each month.

4. While the location of a job is well-defined when someone attends it regularly in person, it can become conceptually unclear in the context of remote work, especially when a job is remote from the start.

5. Metropolitan areas refer to Consolidated Statistical Areas (CSAs) where applicable and Core-Based Statistical Areas (CBSAs) elsewhere. By using the broadest available definitions of metropolitan areas, we minimize situations in which workers separated by arbitrary lines within broader metros, e.g. between the San Francisco and San Jose metro areas, are considered cross-metro. This downside of this approach is that the largest metro areas internalize employment relationships that are cross-metro in kind.

6. The sample is limited to metros with 250,000 residents or more, as observed in the 2017-2021 5-year compilation of the U.S. Census American Community Survey. Smaller places are omitted because their information is likely to be less reliable given the sample’s coverage of workers.

7. Two variants of fitted lines are shown, with and without weighting the fit by metro area population. The results yielded by both variants are consistent with one another throughout the study.

8. The regression using the sample of all U.S. metros with between 250,000 and 1 million residents yields ŷ = -0.23 (0.17) + 0.083x (0.012), with robust standard errors in parentheses. Limiting the sample to metros with more than 1 million residents yields ŷ = -0.27 (0.43) + 0.086x (0.028).

9. Cities with more expensive housing tend to have higher leadership ratios on average as well. However, this relationship is driven entirely by smaller metro areas with populations between 250,000 and 1 million. Among metros with more than 1 million residents, leadership ratios are positively correlated with population but not with typical home values, whereas for those with a population of between 250,000 and 1 million, they are positively correlated with typical home values but not with population.

10. The analysis of changes in metro leadership ratios is less tolerant to data limitations than the analysis of their levels. Because the sample of employers with observed reporting structures is more limited in smaller metro areas, the remainder of the analysis is limited to metros with at least 1 million residents.

11. Because the Zillow Home Value Index (ZHVI) is reported for CBSAs but not CSAs, the log typical home values used are a housing-unit-weighted average of each metro area’s constituent counties’ ZHVI. Throughout the study, the typical home values associated with each metro area were drawn from a single point in time (January 2020, just prior to the pandemic). They were kept static in this way to avoid housing price appreciation from influencing results.

This content was first published by the ADP Research Institute on Nov 8th, 2023.

While hybrid work tensions are getting a lot of attention these days, it is fully remote work that is likely to have greater implications for the economy and its geography in the long run. Although hybrid work reduces the importance of proximity to the workplace and raises tolerance for long commutes, it still requires people to live within commuting range of their workplace.

Fully remote work, on the other hand, requires no such proximity. And what railroads, the shipping container, and globalization did to goods and capital years ago, it can now do to labor. Fully remote work expands the employer labor pool exponentially, offering the promise of cost-cutting and finding perfectly matched talent. It similarly broadens workers’ horizons with respect to their employment opportunities, what they can earn, and where they can live.

Of course, not all the news is good. Fully remote work threatens smaller or regional employers by giving distant, deep-pocketed rivals an opportunity to skim the cream off the crop of local labor. And it exposes workers to a vast pool of distant competition.

While it’s straightforward to observe where fully remote work is being performed, that is, where remote workers live, it’s not clear how to determine where the remotely filled jobs themselves are located, even conceptually. When an on-site job goes remote, the employee’s former desk pinpoints the job’s location, but what if a newly created job is remote? Is the job located at the employer’s site closest to the worker hired? Wherever HR determines? Or perhaps where a worker’s in-person teammates are located, assuming they’re not all remote themselves.

Fortunately, the point is moot. The grand implications of remote work for the economy and its geography do not depend on where the remotely filled jobs are, but on what extent people who live in quite different places are able to work together. What matters is that people work together from different places, regardless of whether the work is done from home, the office, or somewhere else. Moving functions to a back office in a place with cheaper wages was possible well before Zoom, but remote work greatly expands the range of possibilities.

ADP data to the rescue

Collaborating on work from different places is nothing new, but the normalization of remote work during the Covid-19 pandemic has helped take the phenomenon to a whole new level, a shift that ADP data is uniquely well positioned to study. The crucial ingredient for studying the geographic layout of work collaboration is not observing whether an employee spends their days at home or at an office, but rather who works with whom, and where in the nation they are located. We are able to make these observations using ADP data on reporting structures within large firms.[1]

We define teams as groups of employees with a common manager, and we observe whether these workers are all located in the same metropolitan area or distributed across two or more metros.[2] When a worker lives in a different metro area than their manager, we refer to them as a cross-metro worker, and to the work they do as cross-metro work.

This article is the first in a series exploring cross-metropolitan work and putting numbers to it, beginning with its prevalence.

The prevalence of cross-metro work

Prior to the pandemic, the prevalence of cross-metro work was edging up slowly; since then, it has been skyrocketing. As shown in Figure 1, among employees at large firms whose reporting structure is observed, the share of cross-metro workers rose from about 21.9 percent to 22.9 percent from the beginning of 2016 to the pandemic “eve” in February 2020. Since then, it jumped to 31.2 percent as of June 2023.

For employees on distributed teams, which include all members of teams having at least one cross-metro worker, the share of cross-metro workers rose from about 41.4 percent to 42.5 percent over the four years leading up to the pandemic and reached 52.3 percent in June 2023.

Although there is significant overlap between remote work and cross-metro work, the concepts are inherently different, and that distinction helps us understand what we see in the data.

Someone working remotely, for example, from home but in the same metro area as their manager, is performing remote work that is not cross-metro. In that sense, cross-metro work captures less than the full extent of remote work. On the other hand, someone working from an office—or any other work establishment—while reporting to a manager located in a different metro area is performing cross-metro work that is not remote. Workers at a satellite office reporting to a manager at headquarters could fall in this category, for example, as could a local store or factory manager reporting into the regional center. So, in this sense, cross-metro is broader than remote work.

Therefore, it comes as no surprise that the pandemic-driven acceleration in the growth of cross-metro work shows up in almost every industry, from ones that seem remote-friendly, such as finance and insurance or information, to those that do not, such as manufacturing.[3] Similarly, the pandemic-driven acceleration shows up clearly among both workers whose jobs do not require proximity to others, and those whose jobs do.[4]

The demographics of cross-metro workers

In addition to accelerating the growth of cross-metro work, the pandemic also transformed the demographics of those engaged in cross-metro work. Comparing the earnings of cross-metro workers to those of their teammates’ offers a hint of the changes.

Until the pandemic, the wages of cross-metro workers were approximately 1.1 percent higher on average than those of their teammates, i.e., their co-workers who report to the same manager. But as shown in Figure 2, that premium has fallen to just 0.4 percent as of June 2023, converging to near parity.

How is the decline in the cross-metro worker wage premium related to demographics? Before the pandemic, cross-metro workers tended to skew older. Since then, their ranks have seen an influx of younger workers, bringing the group’s composition and average pay closer to that of their same-metro colleagues.

Figure 3 presents the share of cross-metro workers within each age group. The biggest share is found among 45- to 55-year-olds, followed by the 55-65 and 35-45 age groups. Those older than 75 or younger than 25 have the lowest shares working cross-metro.

But that has changed since the pandemic. Figure 4 normalizes the share of cross-metro workers in each age group to a level of 100 on the eve of the pandemic—February 2020—which helps illustrate the growth since then. The 25-35 age group has seen the greatest increase in its cross-metro share. The younger than 25 and 35-45 age groups have seen substantial growth as well.

Indeed, while about 47.8 percent of cross-metro workers were 45 or younger before the pandemic, their share had grown to 53.8 percent by June 2023. In contrast, average pay levels within each of the age groups have essentially remained flat, as shown in Figure 5. The age groups most prevalent among cross-metro workers before the pandemic are, in fact, higher paid on average than the younger workers who have been joining their ranks.

In summary, the pandemic has greatly accelerated the growth of cross-metro work, making it significantly more prevalent than before. It also has changed the composition of cross-metro workers by drawing in more younger workers.

Of course, none of this speaks to the grand implications of cross-metro work for the economy or its geography yet. For more on that, tune in to the next article.

Notes:

1. The sample used consists of a subset of employers in ADP data with at least 1,000 employees in the U.S. Within those firms, the sample was further limited to members of teams with a reliable set of reporting structure fields, as captured by limiting the data to teams in the U.S. in which no more than 10 employees report to the same manager. Although the number varies over time, there are approximately 1.33 million employees that meet these criteria each month on average from January 2016 to June 2023, employed by an average of 3,500 firms each month.

2. Metropolitan areas refer to Consolidated Statistical Areas (CSAs) where applicable and Core-Based Statistical Areas (CBSAs) elsewhere. By using the broadest available definitions of metropolitan areas, we minimize situations in which workers separated by arbitrary lines within broader metros, e.g. between the San Francisco and San Jose metro areas, are considered cross-metro.

3. The share of cross-metro workers in the finance and insurance industry increased from about 18.4 to 21.9 percent from January 2016 to February 2020, and then to 31.1 percent by June 2023. The corresponding figures in the information sector, were 21.3 and 29.0 percent in January 2016 and February 2020, and 43.7 percent in June 2023, indicating much faster growth both before and after the onset of the pandemic. In the manufacturing sector, the share of cross-metro workers increased from about 25.1 to 26.1 percent from January 2016 to February 2020, and then to 31.4 percent by June 2023.

4. Using data on employees’ occupations that is readily available from January 2017, workers were assigned into high- and low-proximity categories based on their occupations’ O*NET scores. A value of 50 was used as the threshold, splitting the 0-100 range in the middle. The share of cross-metro workers in the low-proximity category increased from about 27.2 to 28.6 percent from January 2017 to February 2020, and then to 38.4 percent by June 2023. Although the corresponding figures for the high-proximity group were lower and increased by less after the onset of the pandemic, they exhibit the same general pattern, edging up slowly from about 19.4 to 19.7 percent from January 2017 to February 2020, and then rising more sharply to 23.6 percent by June 2023.

Economic and political power has historically been concentrated in a small number of command-and-control cities, which serve as nerve centers for business, ideas, and influence.

In recent decades, a growing regional divide between costly and affordable U.S. cities has been joined by technological innovations that make it easier for people in different places to collaborate. The combination is shifting low-paying work to more affordable places, leaving value-intensive jobs concentrated in expensive ones. The trend has been many years in the making and was likely further facilitated by the Covid-19 pandemic.

This process, which we call domestic offshoring, is leaving the nation’s command-and-control cities with a distilled workforce, one that’s heavily concentrated with executive decision-makers and other people who are highly paid, while lower-paid workers disperse. This growing economic and geographic divide threatens to limit opportunity for people in lower-cost regions and could have social and political consequences.

To illustrate domestic offshoring, the ADP Research Institute contrasted the whereabouts of top executives with those of customer-service representatives using location quotients, a tool used by economists and geographers to see if a job or industry is more or less common in a particular place compared to the national average.[1] We chose top executives because they are key decision-makers, and customer service representatives because they are the poster child of low-paid but geographically flexible work.

Top executives cluster in expensive cities; customer-service representatives in more affordable ones.

Figure 1A plots the location quotients of executives against typical housing prices for large U.S. metros as of 2021. The line’s upward slope is clear, reflecting the over-representation of top executives in expensive cities.[2] In the San Francisco Bay Area, for example, the location quotient for executives is approximately 140, indicating that these top-level decision makers are 40 percent more prevalent in that part of California than they are nationally. Figure 1B shows the same analysis for customer-service representatives. The downward-sloping trend indicates that these workers are underrepresented in expensive cities.

Domestic offshoring has intensified over time.

As shown in Figures 2A and 2B, cities whose housing prices appreciated the most also had the greatest increase in the prevalence of executives, and the greatest decline in the prevalence of customer-service representatives.

This finding helps substantiate the role of housing price appreciation in driving domestic offshoring. It’s not merely that executives are more prevalent in the nation’s most expensive cities, perhaps owing to their historical legacy as centers of commerce, but also that their prevalence has changed over time in line with the performance of housing markets.[3] For example, Austin, Denver, and Seattle had some of the sharpest increases in the prevalence of executives. Housing prices have risen sharply in these cities, but their economies historically haven’t rivaled those of Los Angeles, San Francisco, or New York.

Population growth, on the other hand, was uncorrelated with changes in the prevalence of top executives or customer-service representatives, further highlighting the role of housing affordability.[4]

The shift was underway well before the Covid-19 pandemic.

Figures 3A and 3B depict fitted lines similar to those in Figures 1A and 1B, but from the years 2005 to 2021. During this timeframe, the upward slope for top executives and downward slope for customer service representatives both grew steeper. That means the over-representation of top executives and the under-representation of customer-service representatives in expensive cities both became more pronounced, aligning with the gradual progression of domestic offshoring.

While the coronavirus pandemic normalized widespread remote work from home, the data show it wasn’t the main driving force behind domestic offshoring. Corporate decisions on where to site call centers, back offices, and other in-person work sites have long played a significant role.

The case of the doctor and the janitor

Not all types of work can be domestically offshored, of course. Tradable goods and services, such as products made in factories and shipped to buyers, are different from non-tradable ones such as haircuts, which must be provided wherever the consumer is. If production and consumption can’t be geographically separated, there should be no evidence of domestic offshoring.

To examine this premise, we considered another pair of occupations, medical doctors and janitors.[5],[6] In contrast to the findings for top executives and customer-service representatives, the changes in location quotients for these two occupations weren’t correlated with housing price growth.[7]

The data show that physicians were more prevalent in the nation’s most expensive cities in 2005, and janitors less prevalent. Over time, however, the differences have evened out, which is inconsistent with them being subject to domestic offshoring.

Where does that leave us?

As technology improves, the scope of work that can be geographically shifted will only expand, and the process of domestic offshoring is likely to grow. But if high-paid and high-powered individuals continue to concentrate in command-and-control cities while others are priced out, regional inequality is likely to widen. Inasmuch as opportunities arise from geographic proximity, domestic offshoring may limit the scope of opportunity available to those living in more affordable parts of the country, while locking them out of the more expensive ones. And that, in turn, is bound to fuel social and political polarization.

Halting the technological progress that is expanding the boundaries of work that can be shifted geographically is neither sensible nor feasible. But undoing the policy constraints that stand in the way of curbing housing price appreciation in the nation’s most expensive places is both sensible and feasible. In other words, domestic offshoring is inherently a symptom of the regional divide in housing costs between the nation’s costly and affordable cities. Avoiding its worst implications requires housing policy to effectively address that divide.

Footnotes

1 An occupation’s location quotient is defined as the ratio of that occupation’s local share of employment and its national share of employment. Here, top executives and synonymous terms correspond to Standard Occupational Classification (SOC) codes beginning with 1110, which also include legislators. Customer service representatives and synonymous terms correspond to SOC occupation code 434051.

2 Underlying data are drawn from the American Community Survey via IPUMS USA (University of Minnesota, www.ipums.org). “Cities” in this analysis refers to Office of Management and Budget’s 2021 combined statistical areas (CSAs) or core-based statistical areas (CBSAs), using CSAs wherever they are available. CSAs are broader and thus prevent interpreting mobility within a large regional metro as domestic offshoring. “Average housing prices” are occupied housing unit weighted averages of county-level Zillow Home Value Index (ZHVI) readings, using occupied housing unit counts from the American Community Survey. ZHVI values are deflated using the consumer price index for all urban consumers (St. Louis FRED series CPIAUCSL).

3 This amounts to using first-differencing to eliminate non-time-varying unobserved error components. Of course, there is no guarantee that cities’ unobserved error components are entirely non-time-varying, or that the functional form of the data generating process is amenable to first-differencing.

4 A linear regression of the change in top executives’ metropolitan location quotients from 2005 to 2021 on those metros’ corresponding percent change in average housing prices yields a slope estimate of 0.215 (0.091), where robust standard errors are reported in parentheses. In contrast, substituting the percent change in population over that period in the right hand side yields a slope of -0.052 (0.272). For customer service representatives the corresponding slope estimates are -0.149 (0.058) with the percent change in housing prices on the right hand side and -0.065 (0.104) when substituting it with the percent change in population.

5 Some physician roles have begun being performed remotely through phone and video appointments and other technologies.

6 Physicians and surgeons refer to SOC codes 291060, 291210 and 291240. Janitors and building cleaners refer to SOC codes beginning with 37201.

7 Changes in physicians’ and janitors’ location quotients have also been uncorrelated with population growth. A linear regression of the change in physicians’ and surgeons’ metropolitan location quotients from 2005 to 2021 on those metros’ corresponding percent change in average housing prices yields a slope estimate of 0.027 (0.080), where robust standard errors are reported in parentheses. In contrast, substituting the percent change in population over that period in the right hand side yields a slope of 0.011 (0.212). For janitors and building cleaners the corresponding slope estimates are -0.009 (0.061) with the percent change in housing prices on the right hand side and -0.084 (0.126) when substituting it with the percent change in population.

It’s well known that in coastal California, homeownership and rents are largely out of reach for low-income earners. But even highly skilled or educated workers earning well above the median income – think nurses, accountants, engineers and even software developers – are being priced out of the housing market in parts of California. 

In an effort to provide California legislators and their staff with accessible and relevant data, we compiled information on the incomes of hundreds of occupations and tracked home prices in regions across the state. What we found were stark regional divides. 

In California’s expensive coastal cities, affordability is beyond reach for low- and middle-income households, and is often a struggle for all but the very highest earners. Further inland, homeownership has remained within reach for those in middle-income occupations, but for those lower down the income spectrum it has become a struggle inland, too.

A straightforward measure of affordability is the price-to-income ratio, which measures typical local home values against annual incomes. A family earning $100,000 a year in an area where a typical house (not necessarily theirs) costs $300,000, for example, has a price-to-income ratio of 3, which economists consider the upper bound of comfortable affordability. An area is considered moderately, unaffordable when the ratio exceeds 3 and severely unaffordable when it exceeds 5. By the time it exceeds 7, homeownership is essentially out of reach.[1]

Statewide, the price-to-income ratio for California’s median resident is 3.9 percent – moderately high. But we’re talking about real estate, so location, of course, is everything. 

Let’s start along the coast, where the housing prices are extreme. In the greater San Francisco region, before the pandemic, the median resident’s annual household income was $117,000, high by most standards.[2] But a typical home at the time cost close to $1 million, putting the price-to-income ratio for the median resident of the region at a suffocating 7.9. Using a price-to-income ratio of 3 as a benchmark, only households earning more than $308,000 a year could comfortably afford to buy a house in the region.

Even highly paid software developers are price-challenged when it comes to homeownership. They’re crucial to keeping Silicon Valley up and running and earn a median $210,000 in the Bay Area. But their price-to-income ratio is 5.8. For registered nurses, whose incomes are lower but who aren’t as concentrated in the most expensive parts of the region as software developers, the ratio is 5.4. And for lawyers, who typically live in higher-income households than software developers, it is still a challenging 4.7.

Price-to-Income Ratio for Median Group Member, 2015-2019

In the greater Los Angeles region, including the Inland Empire, the corresponding numbers are a household income of $78,000 and a price-to-income ratio of 7.1, which imply a household needs to earn almost $185,000 a year to comfortably afford the typical home.

In both the San Francisco and Los Angeles regions, firefighters and police officers appear to punch above their weight with respect to affordability. In greater Los Angeles, for example, those two occupations have higher price-to-income ratios than physicians and surgeons, but lower ratios than lawyers and CEOs, despite substantially lower incomes. 

A likely explanation is that firefighters and policemen often work in multi-day shifts, which make them more amenable to long commutes. In other words, firefighters and policemen can get by living much farther from work than most other people, which allows them to work in one place while living somewhere else that’s distant but much cheaper. 

If it’s a reach for lawyers and software developers to buy, imagine the plight of people farther down the income spectrum. Homeownership in the Bay Area is seriously unaffordable for teachers, architects, electricians, network administrators and accountants, among others. 

For many others, homeownership is simply impossible – a typical house can cost more than 10 times the income of an unskilled worker.

Drive a couple of hours away, however, and the landscape changes. In the San Joaquin Valley, homeownership is within reach for a lot more of the population. Teachers, registered nurses, accountants and, yes, even software developers and lawyers, fall well within the comfortably affordable range. They make less money than they would in the Bay Area, to be sure, but because a typical home can be had for much less, it’s a lot easier to live on their incomes.

But a large share of the population still struggles, even in lower-cost parts of the state. Agricultural workers, who are essential for powering the regional economy, have little chance of buying even in the fields of the Central Valley, where their price-to-income ratio is 6.1. The same holds for skilled laborers in construction trades.

And the problem is getting worse. In every occupation we looked at, in every region of California, wages have failed to keep up with housing prices. 

From 2000 until shortly before the pandemic, household incomes for agriculture workers in the San Joaquin Valley had grown by a third, to $39,000. But the median cost of a home there had more than doubled, to $233,000.

In 2000, roofers, masons and other construction workers could afford a house in the San Joaquin Valley or environs east and north of Sacramento. But wage growth in those occupations over the past 20 years hasn’t kept pace with rising home prices and they’re now priced out.

Price-to-Income Ratio for Median Group Member, 2015-2019

If that trend continues, middle-class earners in dozens of other occupations will soon join their neighbors in the construction trades, falling into the red zone of unaffordability even in communities that today are considered within reach.

The tried-and-true strategy of moving away from job centers to find an affordable home might soon be a thing of the past, too. People willing to accept a long commute in exchange for homeownership might not have that option anymore as sky-high prices spread.

While mounting housing affordability concerns are pervasive in California, they play out differently in every region. Different swaths of society are affected in different places.

The land use policy reforms needed for addressing the problem require coalition-building and cooperation. Hopefully, clear data on housing affordability by occupation in state senate and assembly districts can help find and pull together the coalitions needed for change. 

This blog post uses data from the HOPE Tool, and highlights some of the example insights presented here, as a resource for California legislators and the public.

Notes:

1. These affordability labels are drawn from a related measure, the median multiple, which compares an area’s median household income to its median home price. Here, we are comparing households’ actual incomes with the area’s typical home price (derived from the Zillow Home Value Index). For more information, see here.

2. This income figure and others are drawn from the 2015-2019 five-year compilation of the American Community Survey.

Five decades of economic strength coupled with increasingly restrictive land-use policy have made California unaffordable to most people. It also risks fundamentally altering the divide between haves and have-nots. 

More than ever, existing property owners benefiting from rising home prices stand in stark contrast to recent arrivals to the market, including younger generations of Californians. 

Now there is evidence suggesting that wealth, be it born of homeownership or something else, is gradually eclipsing income as the key factor when it comes to homeownership. The need for wealth has had the insidious dual effect of limiting the pool of buyers and keeping properties held in the family off the market.

Let’s take a step back. To calculate housing affordability, economists typically look at households’ monthly housing costs as a fraction of their income – the housing cost burden. Household incomes have increased over time in California, but housing costs have increased even more. The share of cost-burdened households – those that spend more than 30 percent of their income on housing – has risen for every age group. 

But housing cost burdens consider only monthly costs, such as rents or mortgage payments. They gloss over the need for a down payment or the ability to swap cash for borrowing–when that’s possible. 

Another affordability metric, the price-to-income ratio, takes the full cost of a home into account by weighing the typical local home price against residents’ annual income. By that measure as well, housing in California is unaffordable for most.  

A large cash down payment is one way of getting around high price-to-income ratios and making a monthly mortgage manageable. And just as a large down payment can lower the monthly mortgage by reducing the amount borrowed and lowering the interest rate paid, a small down payment can make the monthly bill much higher.

As home prices soar, so does the size of required down payments. For most people, saving hundreds of thousands of dollars to put down on a house in California can take years or even decades if it can be accomplished at all, especially when soaring rents make it harder to save. 

As savers put off purchasing, the pool of would-be buyers shrinks and changes. Today’s market increasingly favors cash buyers, existing homeowners armed with equity, and people with inter-generational assistance–money from parents or grandparents. 

The latter can be manifested in things like college funds, hand-me-down cars, or a place to crash rent-free for a while. Such assistance seems superficially unrelated to housing and is vastly under-reported as an indirect source of homebuying funds. 

When price-to-income ratios climb high enough, tackling home prices with income alone becomes infeasible, and homeownership comes to depend less on income and more on wealth. In that state of affairs, even very well-earning individuals may find themselves priced out of homeownership unless they’re lucky by birth. 

And if you want your children and grandchildren to live nearby, it could depend on you more than it does on them. 

Take a hypothetical first-time buyer. In greater Los Angeles, adults aged 25 to 34 have a median income of $83,000 and a price-to-income ratio of 7. They’re extremely cost burdened. 

If a typical home costs $587,000, the 20 percent down payment required for the cheapest mortgages would amount to $117,400–more than the young household’s annual income. Closing adds even more to that up-front cost.

For many adults getting married or starting families, homeownership is simply impossible unless they have wealth from day one–or help.

Now a wealth-related phenomenon is emerging. Since 2000, the share of California residents in households who own their homes free and clear has grown across every racial group and among all but the oldest households.

The share of owner-occupied houses with a mortgage, meanwhile, has been shrinking. That’s partly because more people now rent. But it’s also because more residents didn’t take out a mortgage to buy to begin with. They paid cash or inherited their houses.

The growth in free-and-clear ownership can be seen across all regions of the state. The trend extends to even the youngest households, with residents 24 and younger. In addition to reflecting the increasing concentration of the nation’s wealthy in California, this growth might be the result of more young adults living with their parents – or more young adults living in homes their parents bought them.   

Such inherited or gifted houses lock in wealth for future generations, and in some cases keep a property itself in the family for generations. That’s particularly common in California owing to property tax laws – Prop. 13 and its extensions – and the practice keeps a sizable share of homes off the market interminably. 

In the greater Los Angeles region, the share of free-and-clear owners among households with residents aged 24 to 35 – peak household formation years – doubled since 2000 to 8 percent. In the Bay Area, the share more than doubled, from 3 percent to 8 percent.

Population Share by Housing Tenure: California (2000 vs 2015-2019 average)

Conversely, older Californians have suffered a decline in free-and-clear ownership. That’s consistent with today’s prolonged climb up the homeownership ladder. People are becoming first-time homeowners later in life for a variety of reasons, including the greater financial challenge, and mortgages are being paid off later, if at all.

It’s consistent, too, with the theory that more parents are supplying their children directly with down payment help, or indirectly by helping with college or other expenses that would otherwise eat into saving for a down payment. For many parents, one way to help the kids is to borrow against their own homes, which has the side effect of delaying the parents’ own graduation to free and clear ownership. 

In an environment like this, programs that offer low-down-payment loans have limited value. Scrimping on a down payment is always costly. Federal law favors homebuyers with less debt and mortgage lenders give the best terms to borrowers who can put at least 20 percent down.

Even sellers can be biased toward buyers who promise a large down payment, which, in theory at least, reduces the odds that a loan and the sale will fall through.

Today, the homeownership rate in California is 54.2 percent, well below the nationwide rate of 65.4 percent and the lowest of any U.S. state. Only the District of Columbia has a smaller share of homeowners. As the homeownership rate falls, wealth becomes even more concentrated. 

As buying a house gets further out of reach for most Californians, homeownership is becoming less an achievement of the American Dream and more a stark symbol of wealth. 

If California stays on this track, it could eventually come to resemble Manhattan. As of 2019, only 23.3 percent of households in Manhattan owned their homes. In contrast to the rest of the country, homeownership in Manhattan implies a certain class status all on its own.

If you don’t belong to a distinct social class, and you don’t have or expect to have the wealth to buy your kids a home, odds are good that they’ll join the flow of domestic migration from California to cheaper parts of the nation. 

And who can blame them? Five decades of restrictive land-use policy have taken their toll, and that won’t change unless we enact land-use reform that paves the way for enough housing to accommodate everyone who wants to live here. 

That means saying “Yes! We welcome more homes in the neighborhood. Period.”

If you want your children and grandchildren to live nearby, it could depend on you more than it does on them.

This blog post uses data from the HOPE Tool, and highlights some of the example insights presented here, as a resource for California legislators and the public.

The elderly can easily be portrayed as the enviable villains in California’s housing saga. They bought their homes decades ago when they were still affordable, and today they pay disproportionately low property taxes and spend their time fighting new housing, while remaining stubbornly oblivious to the young who are hopelessly slaving away to pay rent.

But as with any caricature that portrayal is unfair. Although there is some truth to it, many older Californians are in fact particularly vulnerable to soaring housing costs.  

Let’s start with the income snapshot. In data spanning 2015 through 2019, people 65 and older had the lowest median household income of any age group in California.. 

Among those aged 65 to 74, some of whom may still be working, the median household income was $73,000, well below the state median of $81,000. For those 75 and older the figure was only $55,000. The income gap between these Californians and those aged 45 to 54 was starker than even the gap between Blacks and Whites.

Median Household Income by Age Group: California (2015-2019 average)

The elderly have always been in a financially precarious position. With their earning years behind them, older Californians have fewer options than other age groups for making more money. And while moving might be an alternative for some graying residents, for many others advanced age can be a roadblock to finding cheaper housing. 

Declining health makes relocating difficult, and downsizing can take a profound emotional toll. A move for any reason, but especially one made under duress, can lead to anxiety, hopelessness, and loneliness, a collection of symptoms some experts call transfer trauma. Moving long-distance to a cheaper part of the country can be especially difficult for the elderly.

Into this mix comes an increasingly common phenomenon, the mortgage- and rent-burdened retiree. The conventional wisdom that most seniors enjoy homes free and clear of monthly payments is false.

In 2000, 37 percent of state residents aged 65 to 74 owned their homes outright. In the years just before the pandemic, 2015-2019 that number had fallen to just 30 percent.  Among people 75 and older, the share of debt-free homeowners fell from 49 percent in 2000 to 42 percent.

Population Share by Housing Tenure: California (2000 vs 2015-2019 average)

Meanwhile, the share of renters among California residents aged 65 and older has grown, reaching 25 percent in 2015-2019. The share of those residents who owned their homes with a mortgage also has grown over that period.

What’s driving the trend? First of all, today’s seniors bought their homes a couple of decades after those who populated the same age groups in 2000. Even if the ramp up of housing prices in the 1970s, ‘80s and ‘90s seems quaint in hindsight, they prevented some people from ever becoming homeowners and they delayed others from starting their mortgage clocks.

The housing collapse that began in 2008 also took its toll. Foreclosures and distressed sales turned many one-time homeowners into renters. By the time property values rebounded, some of those renters were priced out of ownership for good. And others who lost homes during the foreclosure crisis did recover, but had to start over. They took on new mortgages at later ages, which means paying off those mortgages later, too.

For people who held onto their homes, low interest rates and the corresponding refinancing waves extended the life of their loans. The cheap borrowing made for lower payments, but those refis typically reset the clock on 20- and 30-year mortgages that are still being paid off.

Finally, as property values soared, older homeowners tapped their equity with frequency, taking out reverse mortgages or cash-out loans to finance the college tuition–or fund the downpayment on a child’s home purchase.

Seniors who rent are the most vulnerable. The 1 in 4 elderly Californians who rent are caught in the vise of their fixed, lower incomes and rising rents, and are likely spending a larger share of their incomes on shelter than they would have before. Unlike younger residents who can more readily adapt to higher rents, even if that means moving, the elderly often cannot.

What about the seniors who own their homes? Those with a mortgage most often have a monthly cost which is predictable, and all of them have predictable property taxes thanks to California’s infamous Proposition 13 and its extensions. While that 44-year-old legislation has hurt California terribly, it does protect the elderly from housing price shocks to which many of them would be unable to adapt.

Now, California’s elderly population is set to explode. The number of retirees is forecast to grow more than three times faster than the overall population, and the number of residents older than 85 will grow more than four fold by 2060, according to the California Department of Aging. 

That means that our treatment of the elderly will only grow more important. As much as California’s YIMBY’s disdain Proposition 13 and its limits on property taxes, the law has protected the state’s elderly. While tax reform addressing that law and its extensions is sorely needed and long overdue, the risk that California’s home prices will continue rising as they have been requires that any new policy must take into consideration its implications for the vulnerable elderly. After all, a civilization is judged by the way it cares for its helpless.  

This blog post uses data from the HOPE Tool, and highlights some of the example insights presented here, as a resource for California legislators and the public.

In recent years, California has undertaken several unique strategies to better quantify housing needs at the local level and to encourage and require localities to plan for growth. The Housing Element, which requires cities to develop a blueprint for growth at the local level, is one such strategy. The Housing Element is a planning framework that requires all California cities to respond to state targets for expected household growth on an eight-year cycle. For many years, however, these documents have not been truly reflective of subsequent development realities on the ground. Cities may have a certified Housing Element on paper, without much likelihood that the growth specified in the Housing Element would occur, often because unrealistic sites were pegged for development or growth is cordoned off to just one part of a city. Legislators have passed a series of laws over a few years to strengthen the analysis that goes into Housing Element certification. These changes will require more rigorous analysis of how siting new housing is planned. This blog post shows how one city—Los Angeles—used data-driven evidence to justify their Housing Element sites, including original analysis conducted by the Terner Center. Evidence-based Housing Elements may better reflect development conditions on the ground, and may therefore force cities to reconsider their current land use practices in response to the housing goals set by the state. 

Background

A quick primer on the California’s Housing Element process: the State of California’s Department of Housing and Community Development (HCD) assigns each metropolitan region an estimate of the number of units needed over the next 8 years to both meet  existing housing needs and accommodate forecasted population growth. This process is known as the Regional Housing Needs Assessment, or RHNA. Each region then passes along allocations to the individual jurisdictions. Every 8 years, cities must demonstrate that they have adequate capacity of developable sites to meet their share of the regional target, and include this in an updated Housing Element, as required by law. As the draft deadline for the upcoming cycle grows nearer, cities are struggling to meet the new requirements enacted by recent laws. 

Creating realistic Housing Element planning scenarios should now rely on objective estimates of future residential development. The state has adopted increasingly rigorous standards to produce these estimates. One key aspect is that cities may no longer assume that all the residential zoned capacity included in the Housing Element will actually be fully developed in the 8-year cycle. In 2017, AB 1397[1] amended the State Housing Element Law prompting HCD to call on cities to account for the probability of development of non-vacant sites. In the past, cities were allowed to  assume that any parcel that had potential for development would be developed to its zoned capacity within the RHNA cycle with 100 percent certainty. Site capacity estimates were therefore often unrealistic.

More rigorous site standards resulting from AB 1397 have particular implications for California cities that are largely built-out and/or cannot annex or repurpose greenfield development. If those cities cannot justify that parcels with zoned capacity for additional residential use are likely to be developed, then they will be obligated to undertake zoning changes to produce the needed capacity. These upzonings or rezonings can be  politically fraught, time-consuming, and expensive.

In addition to more rigorous site standards, recent reforms to the RHNA process included formula changes[2] that better account for such factors as jobs-to-housing balance, worst-case housing needs, and overcrowding, all of which have been exacerbated by rising rents and home prices. As a result, the City of Los Angeles’ growth targets for the upcoming RHNA cycle have jumped markedly. 

City of Los Angeles takes a new approach.

Recently, the City of Los Angeles released a draft Housing Element incorporating Terner Center Fellow Issi Romem’s estimates relating zoned capacity to future residential development. In collaboration with city staff, Romem developed a model to predict the likelihood and extent of future residential development for each parcel in the Housing Element. To our knowledge, Los Angeles is the first city to include a statistical site inventory analysis of past residential development in their Housing Element. These methods report site capacities grounded in empirical analysis rather than assumption.

In the last RHNA cycle, Los Angeles was obligated to identify capacity for 82,002 new units of housing. In the current cycle, Los Angeles is obligated to identify capacity for 456,643 new units of housing. This top-line increase in targets was a major source of concern to local elected officials when it was preliminarily allocated by the state in March 2021, with threat of possible litigation by cities that felt these new targets would be too hard to hit. Mayor Eric Garcetti, however, pushed the Southern California Association of Governments and the City’s staff to play a constructive role in the process.

“This is a moment of our growing up… I understand the fear where people are like: ‘No, just keep [housing] out and maybe my traffic won’t get worse.’ Well, we’ve tried that for three decades and it’s failed. This is a new beginning.”[3]

– Mayor Garcetti

Facing not only higher overall planning targets, but also a higher bar for estimating site capacities, Los Angeles partnered with the Terner Center to estimate realistic site capacity. As part of the 6th cycle of the Housing Element process, the City of Los Angeles provided HCD with a draft inventory of appropriately-zoned sites to demonstrate that it has adequate capacity to meet its growth targets.

Model Design 

The approach taken to estimate realistic site capacity used recent residential development trends observed in the data as the basis for estimating future residential development. Specifically, it used data on permitting between 2015 and 2019 (5 years) to model the likelihood of new units being permitted on each Housing Element site, as well as the share of zoned capacity that would be developed over the next 5 years. The model accounted for parcels’ zoned capacity at the time with and without bonuses,[4] as well as market conditions and other factors which the city is required to address by law under current requirements.

The model was then applied to current data—including changes over time of parcels’ zoned capacity, market conditions, and other factors—in order to predict the likelihood and extent of future residential development. Crucially, to comply with the requirements of existing Housing Element law (which doesn’t give cities credit for development beyond the bounds of the plans included in their approved Housing Element), the model was constrained so that the predicted number of units on a site could never exceed its capacity (inclusive of bonuses). That constraint is at odds with past development trends: about half of the housing units permitted in the city from 2015 to 2019 exceeded the observed zoned capacity, as a result of reforms to upzone the area during that time period. 

Finally, the approach involved additional consideration of LA’s Transit-Oriented Communities (TOC) program, which significantly relaxed land use policy constraints near transit hubs starting in late 2017.[5]

Findings

The City of Los Angeles’ draft Housing Element marks the first large-scale move towards including data-driven statistical estimates of the likelihood and extent of future residential development in the Housing Element process.[6] The research conducted is also informative in its own right.

The model predicts that, within the bonus-zoned cap, the parcels considered for the city’s Housing Element would see new units amounting to 6.5 percent (29,505 units) of its 456,643 unit allocation permitted within 5 years. Adjusting that number for an 8-year period, such as the RHNA cycle, raises the number to 10.3 percent (47,208 new units), and further adjusting it to account for the TOC program’s benefits brings the number to 13.4 percent (61,158 units).[7]

The likelihood of development of any individual site is low, and high-density multifamily sites are not always developed at full zoned capacity. From 2015 to 2019, about 1 in 80 parcels in the city saw new units permitted each year, which is a 5-year development probability of about 1.2 percent. For the upcoming RHNA cycle, the model predicts that the average parcel will have slightly better odds, with a 1.49 percent development probability over the next 5 years. Even for the most likely sites, the probability of development is fairly low, with the 99th percentile having an estimated development probability of just 8.58.

The vast majority of future new housing units are predicted to occur in multifamily housing, with the median unit being in a 12.4-unit development. If they are developed, sites zoned for 5 or more units are expected to see unit counts typically ranging from 40 to 90 percent of their bonus-zoned capacity, so it cannot be taken for granted that zoned capacity will be used in full. That said, a substantial share of units built in Los Angeles over from 2015 to 2019 occurred in excess of the observed zoned capacity, likely involving changes to entitlement.[8]

As a result, estimates of site capacity are far lower than they would have been otherwise. In fact, Los Angeles already has excess zoned capacity (i.e. zoned capacity above and beyond the existing number of units on a site) that amounts to multiple times its 456,643-unit RHNA allocation. In past cycles, the city could have applied screening criteria to select any arguably defensible subset of sites whose excess zoned capacity added up to 456,643 and met the RHNA allocation by asserting that all zoned capacity on the selected sites would be developed during the cycle, without any burden of proof. Accounting for the estimated likelihoods that sites’ existing zoned capacity will be developed during the cycle—instead of assuming that they will all be developed at full capacity with certainty—during this cycle means that the total estimated site capacity across the city is far lower than it would have been otherwise. The final figure of 44,832 new units over 8 years is less than 10 percent of the 456,643-unit RHNA allocation and so the remaining balance will need to be accounted for through a combination of new capacity from multifamily upzoning, the removal of other barriers to housing production, and complementary production strategies allowed by HCD to supplement the site analysis (such as Accessory Dwelling Unit and public land policies.)

Policy Implications

To our knowledge, no other California cities have undertaken as rigorous an assessment of the probability of development as the City of Los Angeles. Most cities rely on assumptions from planning staff on the likelihood of development. Such assumptions can easily reflect a “planner’s optimism” that development will inexorably flow from favorable zoning and adequately permissive land use regimes alone. While the Terner Center’s approach need not be the only viable method, an analysis with a comparable level of data-driven rigor is necessary for grounding zoned capacity estimates in development realities. To do otherwise would risk defeating the clear language now articulated in state law[9] and ultimately undercut the ability of local governments and the state to understand the actual likelihood that planning targets will translate into new homes being built.

 Yet, it may be a fraught proposition for this level of rigorous modeling to become the norm for the current Housing Element Cycle. Guidance from HCD, which has the administrative authority to clarify expectations for the Housing Element, has been largely non-prescriptive. This reflects the logistical challenge for California’s 540 cities and counties to identify and clean all the relevant parcel-level land use and permit data and run the relevant econometric models. Zoning reform and other regulatory changes that may prove to be necessary through the Housing Element process are also politically complex, and the non-prescriptive nature of HCD’s guidance may be accounting for the limited capacity of locally elected officials and city staff to adopt changes. 

The rolling deadlines for Housing Element Adoption also mean that most California cities have already submitted their draft Housing Elements, creating a large hurdle for implementation if the state were to add new guidance on the Housing Element process. The Bay Area is the only major region that has yet to submit its draft Housing Element and creating a more exacting expectation for compliance at this stage would be criticized as arbitrarily impacting one portion of the state over others. A more nuanced approach may be necessary, such as HCD approval of cities’ 6th cycle Housing Elements conditional on them onboarding this level of analysis for a mid-cycle review four years out, or functionally pushing this provision of law back into the 7th cycle. 

Regardless, investing in better collection and cleaning for parcel-level land use data is a critical next step for local governments, with the support of the state. This level of data collection is necessary for the kinds of analyses that would help local and state policymakers, researchers, and advocates to perform this level of analysis and, more broadly, to understand whether our state housing laws are effective or lag painfully behind. The state, working with and through the regional councils of government, must invest in creating and managing centralized, up-to-date, detailed parcel-level land use data for all California cities.

The sobering reality is that the City of LA can only expect about 0.2 percent of its sites to be redeveloped in any given year. It is a reminder that favorable zoning and land use regimes do not on their own create new housing. Indeed, upzoning, for all its coverage and debate in recent years, is necessary but insufficient on its own for solving the twin challenges of financial feasibility and finite capacity within the housing industry. Building housing in California has become significantly more time-consuming and expensive than it historically was compared to peer regions around the country.[10] This is true when building on an empty site but even more so when redeveloping a site with an existing building. Expanding the housing stock in California’s large metro areas will require addressing both challenges: restrictive land use policy and the high costs of building that housing.

notes

1. See Section 65583.2 (g) (1) For [relevant sites], the city or county shall specify the additional development potential for each site within the planning period and shall provide an explanation of the methodology used to determine the development potential.
2. These new requirements are part of a wave of recent legislation passed to address long standing criticisms to California’s Housing Element Law. For a more detailed dive into criticism on the law itself, RHNA, and how cities estimate site capacity, see the following sources: 

• Ling, S. (2019). “How Fair Is Fair-Share? A Longitudinal Assessment of California’s Housing Element Law.” UCLA. https://escholarship.org/uc/item/45g2k3fp.

• Monkkonen, P., Manville, M., & Friedman, S. (2019). “A Flawed Law: Reforming California’s Housing Element.” https://escholarship.org/uc/item/6dx7914m#main.

• Bromfield, H. & Moore, E. (2017). “Unfair Shares: Racial Disparities and the Regional Housing Needs Allocation Process in the Bay Area.” https://escholarship.org/uc/item/7nx462rf.

• Howle, E. M. (2020). “The State Must Overhaul Its Approach to Affordable Housing Development to Help Relieve Millions of Californians’ Burdensome Housing Costs.” California State Auditor Report, p. 78. Retrieved from: https://www.auditor.ca.gov/reports/2020-108/summary.html.

• Lewis, P. G. (2003). “California’s Housing Element Law: The Issue of Local Noncompliance.” San Francisco: Public Policy Institute Of California. Retrieved from: https://www.ppic.org/content/pubs/report/R_203PLR.pdf.

3. Dillion, L. (2019). “Coastal cities give in to growth. Southern California favors less housing in Inland Empire.” Los Angeles Times. Retrieved from: https://www.latimes.com/california/story/2019-11-07/housing-building-density-zoning-coastal-inland-empire-southern-california-scag.

4. Bonuses can be applied in different ways, e.g. depending on the share of subsidized units and their rent level, so rather than there being a single number that reflects allowed capacity with bonuses there is in reality a menu of such numbers. The numbers provided here were provided by city staff and reflect the city staff’s assumptions as to which bonus structure to apply to each parcel, based on eligibility for a range of affordable housing incentive programs (including Density Bonus, TOC, and Community Plan Implementation Overlay Zones).

5. A full rendition of the model’s methodology, its relation to the legal requirements and its predictive performance is provided in Section 4 of Appendix 4.6 of Los Angeles’ draft Housing Element, and the treatment of the TOC program is provided in Section 5 thereof.

6. For an extensive review of the implications of recently passed legislation on HCD’s power to enforce and reform the Housing Element process, including noncompliance issues for cities unable to meet RNHA goals, look to Chris Elmendorf’s legal review. 

7. City staff subsequently tempered that figure down by manually removing sites whose existing uses would likely preclude development in ways unaccounted for by the model, yielding a final figure of 44,832, or 9.8 percent, of the city’s allocation.

8. As already noted, in order to comply with the requirements of existing Housing Element law (which doesn’t give cities credit for development beyond the bounds of the plans included in their approved Housing Element), the model was constrained so that the predicted number of units on a site could never exceed its capacity (inclusive of bonuses). 

9. See above, AB 1397.

10. Raetz, H., Forscher, T., Kneebone, E., & Reid, C. (2020). “The Hard Costs of Construction: Recent Trends in Labor and Materials Costs for Apartment Buildings in California.” Terner Center for Housing Innovation.  https://ternercenter.berkeley.edu/hard-construction-costs-apartments-california; Reid, C., Napolitano, A., & Stambuk-Torres, B.(2020). “The Costs of Affordable Housing Production: Insights from California’s 9 Percent Low-Income Housing Tax Credit Program.” Terner Center for Housing Innovation at UC Berkeley. Retrieved from:http://ternercenter.berkeley.edu/development-costs-LIHTC-9-percent-california; Reid, C. & Raetz, H. (2018). “Perspectives: Practitioners Weigh in on Drivers of Rising Housing Construction Costs in San Francisco.” Terner Center for Housing Innovation at UC Berkeley. Retrieved from: https://ternercenter.berkeley.edu/uploads/San_Francisco_Construction_Cost_Brief_-_Terner_Center_January_2018.

This content was first published in the New York Times on December 3, 2020, and is reproduced here with permission.

A version of this article appeared in print on Dec. 6, 2020, Section BU, Page 7 of the New York edition with the headline: Housing Market Frenzy Comes With an Explanation.

Ever since the housing market has emerged from its pandemic-driven freeze in late spring, there has been a frenzy of home buying and a sharp rise in prices.

The common explanation is that the pandemic has swollen the ranks of those who need more private space for work and play and safety, typically farther from urban centers.

But why has that translated into fast-rising home prices all around? After all, if buying a new home means leaving behind an old one — an equal contribution to supply and demand — then what’s disturbing the market balance and driving up prices?

A number of factors are at work. Some are well known, and others — like innovations that speed up home buying — have been less noticed and seem as if they may last.

Mortgage rates are low and continue to fall, increasing buying power and helping to fuel housing price growth. But the rates can explain only so much. The current episode of declining mortgage rates began in late 2018 and has been gradual, whereas the current frenzy began in late spring with a jolt.

There is also a relative dearth of homes for sale. Despite a recent increase in new listings that helps compensate for the big drop in the spring, the flow of new listings is still lower than that of previous years.

In addition, the pandemic has increased second-home buying and has crucially brought more first-time buyers into the market.

Typically, 55 percent to 70 percent of American home buyers are selling one home and buying another, with the remainder buying a home for the first time.

Lining up to tour a house for sale in Belleville Township, N.J., in August. Increased demand is part of the story, but not the whole story.Credit...Karsten Moran for The New York Times

Since the spring, the share of first-time buyers in the housing market has risen sharply. Unlike repeat buyers — who contribute equally to demand and supply — first-time buyers contribute only to demand. The share of first-timers, which hovered around 31 percent in 2018 and 2019, increased to about 34 percent during the spring and summer of 2020, according to a National Association of Realtors survey report.

One effect of the pandemic has been to prompt many renters to make the leap into home buying. To some extent, that’s because the U.S. housing market traditionally associates multifamily housing with rentals, and single-family homes with homeownership (despite recent trends to the contrary), leaving those who want more private space fewer options except buying a home.

That accelerated departure from the rental market of home buyers has contributed to a softening of rental prices throughout the nation (as has the consolidation of households as renters move in with friends and relatives). The impact on rents is especially marked in urban areas like San Francisco and New York, where the proximity to jobs and amenities like theaters or museums that normally justifies a high price has lost relevance during the pandemic.

But perhaps the most important factor is the sheer rise in volume of buyers. Adding more buyers and sellers can generate frenzy all on its own, even without altering their balance. Home buyers flock to homes unequally, converging upon the most attractive ones, and in a thicker market that means this subset of homes draws far more buyers than before. (A market grows “thicker” when it has more participants on both sides, while holding their proportion fixed.) The experience of heightened competition around those homes affects perceptions. And when that culminates in higher sales prices, those sales serve as “comparables,” which inform asking prices on subsequent listings (including less attractive ones), lifting housing prices more quickly.

But another explanation for the current buying boom has so far largely escaped attention. Thanks to innovation, the housing market is operating faster than it used to, and that has helped make supply and demand imbalances more extreme.

Here’s how it works. Suppose that for every two sellers in the market, there are three buyers. In other words, the market is a hot one, dominated by sellers. Accelerating the pace at which buyers and sellers successfully pair up and exit the market reduces the number of buyers and sellers equally, and can be thought of as leaving two buyers for every one seller.

Because that ratio of two to one is even more unbalanced than three to two, it implies more intense competition over homes for sale. (The opposite would be true in a buyers’ market, in which faster matching of buyers and sellers would make buyers scarcer than before.) By helping speed up the home-buying process, innovations also reduce for-sale inventory and shorten the length of time homes are on the market. (In isolation, it also shortens the time buyers are in the market.)

Just as technology for remote work has been available for years — but has only now come into wide use because of the pandemic — innovations in home sale transactions have been widely embraced only recently despite being around for a while.

Those include substitutes for in-person visits like 3-D home tours, drone footage and virtual staging that lets viewers imagine how a space could be used. The innovations also include mundane but meaningful things like phone-based entry into homes for sale (no appointment necessary), or the availability of remote notary services, which streamline the transaction process.

The key here is that such innovations are likely to make home buying faster. Showing a home virtually, for example, eliminates the delay imposed by the need for a weekend open house, and a remote online notary can save time when local ones are in short supply. Again, it’s not that these innovations are necessarily new, but that they are now being taken up much more broadly than before.

All of these factors contribute to the buying frenzy, but some are likely to outlast others. The pursuit of private space because of health fears may last a good while, perhaps even for a few years, and a low-interest environment even longer. But neither is likely to be permanent.

If the pursuit of private space is driven by remote work and the need for office space in the home, there is a greater chance it will persist (though not at the shelter-in-place extreme).

But to the extent that innovations accelerating the home sale process are the culprit for a more extreme supply and demand imbalance, well, there’s no turning back the clock on that.

This content was first published in the New York Times on April 20, 2020, and is reproduced here with permission.

A version of this article appeared in print on April 21, 2020, Section B, Page 5 of the New York edition with the headline: A New Kind of Homeownership Could Gain Traction.

We take for granted that homes are either owned or rented, but that could soon change, and pandemic-driven economic pain could accelerate the process.

Before the pandemic, homeownership was already slipping out of reach for many in America’s expensive coastal cities. Hybrid arrangements have been emerging that occupy a middle ground between owning and renting — trying to combine the advantages of both.

It’s an idea that could bring the benefits of owning a home back within reach, especially for millennials struggling to get on the homeownership ladder.

It’s also an idea that could gain momentum during the pandemic. As economic hardship deepens, investment capital could help desperate homeowners avoid selling their homes while providing investors with the bargain of a century.

But hidden within the idea is a danger: It might work far better early on than in the long run.

Here’s how it works

Instead of just lending people money to buy homes, companies are now co-investing with them — in other words, taking an ownership stake in the home. This can take on many different forms, including offering down payment assistance or providing a substitute for home equity loans.

The underlying principle is the same: A home buyer or existing homeowner receives cash in exchange for a share of the ownership of the home. The co-investor then (1) shares in the home’s rising or falling value (often disproportionately to their ownership), and (2) sometimes requires a flow of monthly payments from the occupant.

When monthly payments are required, they are akin to dividends on the co-investor’s share in the home, and they also conveniently resemble both rent payments and interest payments on a mortgage. In some cases, a portion of the payment is allocated toward increasing the occupant’s equity in the home, akin to principal payments on a mortgage but governed by different math.

For the time being, this arrangement is mostly happening alongside traditional financing of the home. In other words, companies are taking an ownership stake in homes bought with cash or with a mortgage.

But at least one firm, San Francisco-based Haus, is already co-investing most of a home’s value, eliminating the need for a mortgage altogether. It requires monthly payments that can dial up or down the occupant’s ownership share, or keep it steady. And it allows people to roll over the co-investment with new terms at 10-year intervals. (The current practice is to require people to sell or refinance the home after several years so the co-investor can cash out).

Setups in which the occupant never attains full ownership amount to a new way of financing shelter — an owner-renter hybrid. The occupant of the home, without fully owning the property, simultaneously enjoys some of the benefits of homeownership and rental living. Specifically, occupants have some of the security of a homeowner. They need not worry about the landlord’s renewal of the rental contract or about rising rents for an extended period — nor must they come up with funds to ultimately acquire ownership, such as principal payments on a mortgage.

The idea of an owner-renter hybrid isn’t entirely novel. It resembles rent-to-own arrangements, including those of some high-profile new start-ups, as well as some more esoteric financial arrangements for shelter that have cropped up in the past, such as shared appreciation mortgages. In fact, if occupants keep their ownership share steady, it achieves the same end as a very long-term rental contract.

This hybrid also has much in common with rent control. Like rent control, it guarantees a predictable flow of housing costs for the long run and affords the resident protections — from eviction, for example — that provide a degree of permanence.

That’s no coincidence. The recent resurgence of rent control and the own-rent hybrid both address people’s need to replace homeownership with more affordable forms of secure shelter.

Things could get ugly

Enter the pandemic, and the owner-renter hybrid may suddenly appeal to people well beyond the expensive coastal cities. Many homeowners without work may have to sell their homes, even on unfavorable terms. The hybrid arrangement would grant such a homeowner the option of selling equity in the home piecemeal to co-investors for cash, while continuing to reside in it. For many, that would be a much better outcome than the alternative.

It’s also one that would leave investors pleased, with a previously unimaginable stake in owner-occupied housing, acquired on favorable terms. If investors execute their business wisely, they could even pose as saviors, having offered desperate homeowners a lifeline that spared them from catastrophe. But can any such profiting in the aftermath of a pandemic avoid ultimately being judged as predatory? The hybrid arrangement could easily devolve into taking advantage of homeowners in distress, and even the most ethical investors may feel compelled to participate for fear of missing out on a land grab.

Easing the pressure on some buyers

Fast forward past the pandemic and its aftermath: Is the new hybrid the solution for the housing affordability crisis in America’s expensive coastal cities?

It could certainly help ease the pressure for potential buyers. As long as most people prefer to purchase homes outright, principal and all, those willing to go the hybrid route will have an advantage. They will need to pay only interest-like dividends to their co-investors, without having to buy out the co-investor’s share — the equivalent of not paying mortgage principal.

You might wonder: Wouldn’t people lose out by not participating in America’s biggest forced-saving plan ever, the mortgage? The answer is probably yes, inasmuch as the forced commitment to save is the key.

Nevertheless, fans of the new arrangement would justifiably argue that hybrid owner-renters could invest their savings relative to mortgage costs in whatever they choose, including the acquisition of other residential real estate (for example, in the form of tiny fractions of many other homes).

But that’s not how things are likely to play out. Instead of investing those savings, people might be tempted to divert them into greater housing expenditure, by securing more expensive homes than they could otherwise afford to buy.

Not addressing the fundamental problem

It’s not just about individuals’ choices. A mass participation in owner-renter hybrids in the expensive coastal cities would have unintended consequences for the way in which home prices evolve, and for broad housing affordability.

It’s helpful to understand that these hybrids essentially increase the demand for housing. When owner-renters use their savings relative to mortgage costs to obtain better homes, they squeeze more buying power out of their monthly housing budgets by drawing greater amounts of investors’ capital into the housing market. That means more money going after housing.

If owner-renter hybrids take hold, we should expect Wall Street to own a bigger slice of America’s neighborhoods (not just rentals). And we should expect home prices to gradually shift from familiar momentum-driven patterns — based on traditional markets of home trading among people — to the more erratic and volatile behavior of financial markets.

The savings relative to mortgages from the hybrid approach will generally be used for obtaining better housing. People tend to want to buy the best home they can afford — whether that means bigger, nicer or better located — and are usually willing to commit to it their full financial might. In fact, the appeal (and curse) of a mortgage is that it allows people to channel into a home purchase their earning capacity over the next 30 years.

If people were so eager to get the best home they could afford — to the point that they all opted to play the hybrid card to squeeze the most buying power out of their monthly housing budgets — what would that do to housing prices?

It would bid them up higher.

How much higher? Economists would tell you that it depends on how easy it is to build more housing (they would call it “housing supply elasticity”). If increased demand caused developers to build more, prices might rise only modestly, and this might well be the case in some parts of the country.

But if developers couldn’t build much more, as is the case in the expensive coastal cities today, housing prices could rise substantially.

That rise could altogether negate the buying-power gains of those adopting the owner-renter hybrid, and at that point they would be no better off in terms of affordability. Those trying to buy the old-fashioned way would be worse off.

More generally, increasing the demand for housing cannot resolve the public’s housing affordability woes in America’s expensive coastal cities. Only increasing supply can. Housing costs are high in those cities because for decades strong demand for housing has been met with local land-use policy that severely limited new construction. Even a pandemic, with a recession in its wake, won’t stem housing price appreciation in the long term.

Increasing people’s buying power, even if it emerges from something as significant as giving up on traditional ownership, won’t improve the state of housing affordability. As long as demand is fundamentally strong, building more housing is the only thing that will help.