To address this, we’ll need to find ways to increase resources directed at lifelong learning. The chart below shows how education and training spending is extremely front-loaded within the average lifetime. That model suggests people can learn most of the skills they need when they are young and then just apply them throughout the rest of their life. But that doesn’t sound like a successful strategy for our current economy, much less in the future.

As I highlighted last week, there are many companies looking to help tackle parts of this problem.

1️⃣ Employers rethinking how they help workers reskill and upskill
2️⃣ New adult education providers focused on career switchers
3️⃣ New approaches to outplacement and career navigation

But these approaches are quite different—and not equally promising for workers, investors, or society. Some of these approaches suggest a hopeful idea that employers will be a big part of the solution. To understand how realistic that is, it’s helpful to get a clear understanding of employers’ incentives to invest in upskilling and reskilling and identify why they aren’t doing more already. (To clarify, this entire discussion excludes situations where there is a clear career progression that an employee can take advantage of over time, either through becoming more high-skilled in the same role or moving into management within the same function).

Employers have strong incentives to upskill and reskill their workers when the benefits of doing so exceed the costs of doing so. That’s a simplistic but a helpful starting point.

The costs are straightforward—they are just the incremental cost of the relevant training or education program.

The benefits (focusing only on the employer’s perspective here) can take several forms: The holy grail of reskilling is taking a worker in an entry-level, low-skill job and equipping them for a higher-skill, in-demand and hard-to-fill role at the same employer. A more moderate version of this would be increasing the productivity of the employee in the same or similar role.

This sounds fantastic—the employer addresses a labor shortage, the employee gets more skills and a higher paying job. Win-win.

But in practice the incentives usually don’t line up quite so neatly. In order for the employer to benefit from this, they need to save enough costs from this arrangement versus hiring externally to justify the education costs. To benefit more than just in saving one-time hiring costs, they need to pay the upskilled worker less than they would pay to hire someone for the role externally. But in that case, the employee would have a strong incentive to look elsewhere for a role at a different employer paying the market rate for those newly acquired skills. If the employer plans to pay the upskilled employee the market rate instead, there isn’t much benefit for the employer, which undermines the value proposition of paying for the upskilling in the first place. This can be partly overcome with employee obligations to stay at the company for a certain period after completing their training or education, but those have their own challenges from whether employees will be willing to accept them as part of a program to whether the company is willing to enforce them (I.e., suing a former employee who leaves early). Vesting schedules for things like stock options can also mitigate this effect but are rarely in the picture for the types of roles being discussed.

Other factors at play are also not trivial. The employees who make promising candidates for upskilling need to be located in the same geography that the employer’s high-demand roles are or be willing to move (which is relatively unlikely given the low geographic mobility of Americans today). For many employers, their existing workers are spread out in different areas from where their in-demand roles are. More importantly, not all employers have vast numbers of in-demand high-skilled job demands, even before the pandemic.

Risk and uncertainty also create problems for the cost-benefit tradeoffs involved. Many companies don’t have a high-confidence view of their workforce needs in three or four years’ time. A recurring theme at Remixing Work is the high degree of change in the broader economy and in industry structure. A company’s uncertainty about these factors going forward, as well as uncertainty around its own strategy and performance all flow through to uncertainty about future workforce needs. This is a significant problem in this context because the costs are effectively guaranteed, only the benefits are uncertain.

Sometimes the benefits are high enough, the uncertainty is low enough, and the issue of retaining employees can be overcome. But as a result of these complications, the incentives generally work against significant employer spending for upskilling along these lines.

There are some alternative rationales that justify employer spending in certain cases that I will come to in a moment, but it’s important to emphasize that the above discussion only shows that it will not make financial sense for employers to spend substantially on upskilling low-skilled workers in many cases. However, that does not at all reflect the cost-benefit trade-off from society’s perspective or from the workers. But we shouldn’t look to employers as the primary solution for upskilling (unless we change their incentives through laws or regulation).

So, what about the alternative rationales I mentioned? The most promising of these is illustrated with an example I highlighted previously: the major employers who are providing their employees education as a benefit through Guild Education.

Guild has primarily built their early business around a totally different logic for employer-funded education They have focused on employers in industries with very high employee turnover. Looking at pre-Covid-19 conditions, turnover for hourly workers in retail is commonly in the 50-75% range, and, at fast-food restaurants, it can be 130-150%. This means the typical employee stays less than a year, which implies a huge amount of time dedicated to hiring and training workers. But employees who enroll in employer-sponsored or subsidized education programs are likely to stay years instead of just months. This truly is a win-win and can make a real contribution to upskilling employees. At the same time, it’s important to realize that the actual education is somewhat besides the point from the employer’s perspective-the main benefit is the increased retention. More importantly, this logic only works for some employers in some situations. Employers that don’t face high turnover won’t have the same incentives.

Training our workforce for a changing economy is an important challenge. But we shouldn’t expect employers to solve the workforce’s upskilling needs under the current incentives. There are interesting models that could be considered, for instance, creating incentives for apprenticeship-style programs for career switchers within an employer. Furthermore, startups looking for a market opportunity in the current skills gap will need a clever orthogonal model like Guild’s or a very targeted model to succeed.

But we should also look beyond employers to support and encourage workers more directly. There is a huge opportunity to better serve career switchers and other adult learners directly with education offerings. Part of the solution is new education programs different from what are widely available now. But a big part also needs to be better support for working adults to be able to take advantage of education programs. Innovative financing like ISAs can significantly help. More supportive government regulation and financial support may also be appropriate.

More than one-third of Americans would look to change careers if they lost their job due to COVID-19, according to recent polling (see below). A similar proportion of Americans believe they would need more education to replace a lost job. COVID-19 may be the immediate cause of job losses in these cases, but the need for many Americans to switch careers or get more education to stay competitive in the workforce is not new or specific to the current crisis. Employers and talent consultants have also been highlighting the "skills gap" for years.

"If I lost my job, I would look to change careers"‌

‌Source: Strada Education

"If I lost my job, I believe I would need more education to replace it"‌

‌Source: Strada Education

(Note: There is likely overlap between the respondents answering yes to both questions but that data is not available).

Unfortunately, even in good times, the labor market is not easy for career switchers to navigate. Moreover, traditional higher education offerings have a mixed record putting adult learners into better jobs.

However, there are promising innovators that can help address the immediate jobs crisis related to the pandemic and the longer-term skills gap:

• 1️⃣  Employers rethinking how they help workers reskill and upskill
• 2️⃣  New adult education providers focused on career switchers
• 3️⃣  New approaches to outplacement and career navigation

The innovators

1️⃣  Employer-led reskilling & upskilling

Upskilling and reskilling programs are primarily designed to fulfill an employer's future demand for higher skilled jobs by training for some of its existing employees for new skills. This can help address gaps in high-demand, high skills jobs that are more and more expensive to fill for employers. It also helps provide more pathways for employees at companies with large unskilled workforces but limited advancement opportunities and at companies that anticipate reductions in the future (e.g., due to automation).

Corporate learning & development initiatives have always addressed these needs to a limited extent, but the growing skills gap brought increased attention to this before the pandemic.

One set of companies is trying to make employee's skills more legible and then make it easier to match them with relevant potential future roles along with appropriate training. Degreed is a notable startup in the space. It's customers are mainly large corporates, such as Unilever, Cisco, AthenaHealth, AirBnb, Boeing, Ericsson and Booking.com (all were reported customers as of June 2019). Degreed catalogs the existing skills of employees, helps them identify roles that could be suitable for the future and matches them with training resources to gain those skills. To recognize these skills, Degreed offers more than 1,5000 certificates and credentials. Founded in 2012, Degreed has raised $182M, including a recent round of $32M in June.

Other education providers are also focusing on upskilling corporate employees. MOOC providers Coursera and Udacity launched as consumer-focused learning platforms, but have been focused on growing the increasing share of revenue they get from enterprise customers.

Guild Education is one of the fastest growing education startups and partly plays in this space. They help large employers offer "education as a benefit" for their employees. This is an evolution of the tuition assistance programs that companies have offered for decades. Guild's core customers (companies like Wal-Mart and Chipotle) are large employers in high-turnover industries. By offering free or subsidized degree programs at Guild's partner colleges and universities, they are able to significantly improve employee retention.  This model has tremendous promise for helping employees complete or gain new degrees with little cost to them.

2️⃣  New career-focused adult education providers

Newer online education providers are also catering specifically to career-switchers by emphasizing job placement in their program design and in the financing they offer students.

Lambda School offers 9-month and 18-month programs in web development and data science, Lambda is a pioneer in the use of income-share agreements in which students are only responsible for paying back their tuition after they have a job in their field of study paying at least $50K in salary. As I've noted previously, these funding models align school and student incentives much more than traditional college financing, including for Lambda to put a heavy focus on job placement for graduates. Founded in 2016, Lambda told EdSurge it has ~3,000 students enrolled and has raised $122M including $74M in August

Other online education startups offer career-switching focused programs in similar fields, as well as in web & app design, sales and more. Most also have a strong career placement focus. Many also offer ISAs or provide some kind of tuition reimbursement guarantee.

3️⃣  Better navigation

Software like Degreed helps companies manage career paths within their companies. Career-switching employees also need tools to help them navigate the labor market beyond their employer.

Career Karma is a promising navigation offering that helps people transition to jobs in technology. Its app and community of peers and mentors helps career switchers decide if what type of role in technology would be a good fit for them and then matches them with a bootcamp program. It's free for students--Credit Karma gets a placement fee for the bootcamps it partners with. That revenue model leaves some room for misaligned incentives between Credit Karma and students, but the program gets great reviews and works with a wide-range of top bootcamps (so it's not just steering students to a small number of partners). The biggest limitation to the model for now is that it's focused on a set of specific tech roles. The Career Karma web site provides advice on a broader set of career options but is not designed to pair students with education in those areas at this time (or help them navigate those careers in the same way).

Boost by Kaplan is a new career navigation program for high school graduates. Targeted especially at students taking a gap year before college and just in the process of being launched, it aims to give students exposure to professional work experience and a range of careers before they start college. The marketing copy nods to the idea that students may want to skip college completely after the program and go straight into a promising job (but it's not explicit on that point). Even if most students do go to college anyway after a gap year, they will be in a much better position to make decisions throughout college that prepare them for the workforce having gotten more early exposure.

Two employer-led initiatives also merit note here in the context of navigation:

• Guild (discussed above) has launched a new service called Next Chapter that is reinventing outplacement services for large companies that are laying off workers during the current pandemic. Called Next Chapter, it aims to pair laid-off workers with relevant education as part of their outplacement benefits and also connect them with relevant roles at companies in its employer partner network that hiring. This isn't entirely employee-centered (given the focus on Guild's employer and academic partners) but it still may be a big improvement for workers versus traditional outplacement, especially in a time of crisis.
• Amazon has a major workforce development program that aims to train workers for higher skilled jobs at Amazon but also for higher skilled jobs outside of Amazon. It's designed with outside jobs in mind because Amazon's warehouses have far more entry-level jobs than higher-skilled jobs. Like other employer education programs, it likely has significant retention benefits for Amazon. But what distinguishes it and merits mentioning here is the strategic (albeit mildly paternalistic) selection of education offerings it offers. Specifically, it only includes programs focused on skills that it expects will be in high-demand in the local region and that pay more than Amazon's $15 per hour minimum wage. By filtering it's offering in this way, it automatically provides a kind of navigation service to its employees.

The outlook

Even before the pandemic, the skills gap indicated a strong need for training, education and navigation solutions to help workers and employers adapt. Between the job losses and accelerated economic change caused by the pandemic, the need for these solutions is even more urgent.

Each of these areas offers the potential to create substantial opportunity for workers and to build big businesses. However that opportunity looks quite different by area:

• Career & education navigation: The potential value unlocked by better career and education navigation is tremendous. But the feedback cycle to prove their effectiveness is incredibly long, and they may create the most value during life stages that it is more difficult to monetize (for instance, before or during college). But programs like Career Karma that target a shorter time horizon and are attached to a purchase (either education purchase or job placement) offer a promising path to adoption.
• Targeted adult-education: Targeted vocational programs that serve job switchers well will have the most long-term opportunity. If they not only help career switchers successfully switch but also turbocharge those new careers, they have a clear path to sharing in that value creation. Their biggest challenges are the many forces working in favor of higher ed incumbents and the relatively long feedback cycle involved in proving their effectiveness.
• Corporate reskilling & upskilling: The typical examples make these sound like no brainers for employers, and good for employees and society, too. In practice, they require an unlikely confluence of factors to be a slam dunk for employers. Nevertheless, corporate learning & development is a very large market ($169B in North America), so there is plenty of room for providers targeting this with a reskilling or upskilling focus to be large businesse, too.

Remixing Work is for the entrepreneurs, executives and educators shaping the future of work and education

First, thank you to everyone who shared the last edition and a special welcome to our new readers--we've grown another 25% over last week.

If someone forwarded this to you, you can subscribe here.

I'm working on adding comments to the web site, and they should be up sometime next week. But in the meantime it's always great to hear what you think!

More than one-third of Americans would look to change careers if they lost their job due to COVID-19, according to recent polling (see below). A similar proportion of Americans believe they would need more education to replace a lost job. COVID-19 may be the immediate cause of job losses in these cases, but the need for many Americans to switch careers or get more education to stay competitive in the workforce is not new or specific to the current crisis. Employers and talent consultants have also been highlighting the "skills gap" for years.

"If I lost my job, I would look to change careers"

Source: Strada Education

"If I lost my job, I believe I would need more education to replace it"

Source: Strada Education

(Note: There is likely overlap between the respondents answering yes to both questions but that data is not available).

Unfortunately, even in good times, the labor market is not easy for career switchers to navigate. Moreover, traditional higher education offerings have a mixed record putting adult learners into better jobs.

However, there are promising innovators that can help address the immediate jobs crisis related to the pandemic and the longer-term skills gap:

• 1️⃣  Employers rethinking how they help workers reskill and upskill
• 2️⃣  New adult education providers focused on career switchers
• 3️⃣  New approaches to outplacement and career navigation

The innovators

1️⃣  Employer-led reskilling & upskilling

Upskilling and reskilling programs are primarily designed to fulfill an employer's future demand for higher skilled jobs by training for some of its existing employees for new skills. This can help address gaps in high-demand, high skills jobs that are more and more expensive to fill for employers. It also helps provide more pathways for employees at companies with large unskilled workforces but limited advancement opportunities and at companies that anticipate reductions in the future (e.g., due to automation).

Corporate learning & development initiatives have always addressed these needs to a limited extent, but the growing skills gap brought increased attention to this before the pandemic.

One set of companies is trying to make employee's skills more legible and then make it easier to match them with relevant potential future roles along with appropriate training. Degreed is a notable startup in the space. It's customers are mainly large corporates, such as Unilever, Cisco, AthenaHealth, AirBnb, Boeing, Ericsson and Booking.com (all were reported customers as of June 2019). Degreed catalogs the existing skills of employees, helps them identify roles that could be suitable for the future and matches them with training resources to gain those skills. To recognize these skills, Degreed offers more than 1,5000 certificates and credentials. Founded in 2012, Degreed has raised $182M, including a recent round of $32M in June.

Other education providers are also focusing on upskilling corporate employees. MOOC providers Coursera and Udacity launched as consumer-focused learning platforms, but have been focused on growing the increasing share of revenue they get from enterprise customers.

Guild Education is one of the fastest growing education startups and partly plays in this space. They help large employers offer "education as a benefit" for their employees. This is an evolution of the tuition assistance programs that companies have offered for decades. Guild's core customers (companies like Wal-Mart and Chipotle) are large employers in high-turnover industries. By offering free or subsidized degree programs at Guild's partner colleges and universities, they are able to significantly improve employee retention.  This model has tremendous promise for helping employees complete or gain new degrees with little cost to them.

2️⃣  New career-focused adult education providers

Newer online education providers are also catering specifically to career-switchers by emphasizing job placement in their program design and in the financing they offer students.  

Lambda School offers 9-month and 18-month programs in web development and data science, Lambda is a pioneer in the use of income-share agreements in which students are only responsible for paying back their tuition after they have a job in their field of study paying at least $50K in salary. As I've noted previously, these funding models align school and student incentives much more than traditional college financing, including for Lambda to put a heavy focus on job placement for graduates. Founded in 2016, Lambda told EdSurge it has ~3,000 students enrolled and has raised $122M including $74M in August

Other online education startups offer career-switching focused programs in similar fields, as well as in web & app design, sales and more. Most also have a strong career placement focus. Many also offer ISAs or provide some kind of tuition reimbursement guarantee.

3️⃣  Better navigation

Software like Degreed helps companies manage career paths within their companies. Career-switching employees also need tools to help them navigate the labor market beyond their employer.

Career Karma is a promising navigation offering that helps people transition to jobs in technology. Its app and community of peers and mentors helps career switchers decide if what type of role in technology would be a good fit for them and then matches them with a bootcamp program. It's free for students--Credit Karma gets a placement fee for the bootcamps it partners with. That revenue model leaves some room for misaligned incentives between Credit Karma and students, but the program gets great reviews and works with a wide-range of top bootcamps (so it's not just steering students to a small number of partners). The biggest limitation to the model for now is that it's focused on a set of specific tech roles. The Career Karma web site provides advice on a broader set of career options but is not designed to pair students with education in those areas at this time (or help them navigate those careers in the same way).

Boost by Kaplan is a new career navigation program for high school graduates. Targeted especially at students taking a gap year before college and just in the process of being launched, it aims to give students exposure to professional work experience and a range of careers before they start college. The marketing copy nods to the idea that students may want to skip college completely after the program and go straight into a promising job (but it's not explicit on that point). Even if most students do go to college anyway after a gap year, they will be in a much better position to make decisions throughout college that prepare them for the workforce having gotten more early exposure.

Two employer-led initiatives also merit note here in the context of navigation:

• Guild (discussed above) has launched a new service called Next Chapter that is reinventing outplacement services for large companies that are laying off workers during the current pandemic. Called Next Chapter, it aims to pair laid-off workers with relevant education as part of their outplacement benefits and also connect them with relevant roles at companies in its employer partner network that hiring. This isn't entirely employee-centered (given the focus on Guild's employer and academic partners) but it still may be a big improvement for workers versus traditional outplacement, especially in a time of crisis.
• Amazon has a major workforce development program that aims to train workers for higher skilled jobs at Amazon but also for higher skilled jobs outside of Amazon. It's designed with outside jobs in mind because Amazon's warehouses have far more entry-level jobs than higher-skilled jobs. Like other employer education programs, it likely has significant retention benefits for Amazon. But what distinguishes it and merits mentioning here is the strategic (albeit mildly paternalistic) selection of education offerings it offers. Specifically, it only includes programs focused on skills that it expects will be in high-demand in the local region and that pay more than Amazon's $15 per hour minimum wage. By filtering it's offering in this way, it automatically provides a kind of navigation service to its employees.

The outlook

Even before the pandemic, the skills gap indicated a strong need for training, education and navigation solutions to help workers and employers adapt. Between the job losses and accelerated economic change caused by the pandemic, the need for these solutions is even more urgent.

Each of these areas offers the potential to create substantial opportunity for workers and to build big businesses. However that opportunity looks quite different by area:

• Career & education navigation: The potential value unlocked by better career and education navigation is tremendous. But the feedback cycle to prove their effectiveness is incredibly long, and they may create the most value during life stages that it is more difficult to monetize (for instance, before or during college). But programs like Career Karma that target a shorter time horizon and are attached to a purchase (either education purchase or job placement) offer a promising path to adoption.
• Targeted adult-education: Targeted vocational programs that serve job switchers well will have the most long-term opportunity. If they not only help career switchers successfully switch but also turbocharge those new careers, they have a clear path to sharing in that value creation. Their biggest challenges are the many forces working in favor of higher ed incumbents and the relatively long feedback cycle involved in proving their effectiveness.
• Corporate reskilling & upskilling: The typical examples make these sound like no brainers for employers, and good for employees and society, too. In practice, they require an unlikely confluence of factors to be a slam dunk for employers. Nevertheless, corporate learning & development is a very large market ($169B in North America), so there is plenty of room for providers targeting this with a reskilling or upskilling focus to be large businesse, too.

Until next week!

To bring attention to the potential long-term effects and to build the case for taking additional mitigation actions, a group of World Bank economists estimated the potential long-term impact of the COVID-19 pandemic quantitatively. They published their analysis in a report in May but have been promoting the findings more recently. In a summary posted on Brookings at the end of July they write:

The students currently in school stand to lose $10 trillion in labor earnings over their work life. To get a sense of the magnitude, this sum is one-tenth of global GDP, or half of the annual economic output of the United States, or twice the global annual public expenditure on primary and secondary education.

The report's framing around their $10-trillion estimate is attention-grabbing but mostly unhelpful--it's not tied to actionable policy options and the uncertainty involved is huge and not sufficiently characterized. Beyond the top-line number, the report identifies several negative short-term and long-term effects the pandemic will have on students around the world.

In applying the standard models for assessing the returns of education to the shock of the pandemic, the report also casts some of the major assumptions of such models in a new light. These assumptions are deeply embedded in policy thinking about education. They likely also appear in your own views about education at times. But these assumptions are often inappropriate and can be quite limiting. I will share a bit more about the report and then return to this point below.

The report identifies three main impacts of school closures:

• Increased dropout rates
• Missed learning for the time school is closed
• Increased learning loss from the longer period out of school

The increased dropout rates stand out as the most concerning and non-obvious effect because students who drop out miss the schooling they would have received during the closures (like all students) and miss the subsequent period they would have remained in school if there wasn't the disruption.

The other two effects are notable, but they also bring to light deeply ingrained assumptions about how education translates to learning and how learning translates to higher earnings.

The report's main model is built around a metric called learning-adjusted years of schooling (LAYS). LAYS is a country-level measure of the number of years of schooling that students in that country complete on average, adjusted by a quality factor for that country. To estimate the impact of school disruptions, the report calculates the pandemic's effect on LAYS. It then translates that into a change in earnings using returns to education estimated elsewhere.

The LAYS  metric was developed and published in 2018 by several other World Bank economists to replace the metric previously used for cross-country comparisons, which only considered years of schooling without any quality adjustment. Except in its use of the s newer quality-adjusted metric, the report's approach using years of schooling as the primary input variable has been the standard methodology in labor economics for nearly 50 years.

The origin of using years of schooling to explain variation in earnings was the ground-breaking empirical work of Jacob Mincer. Mincer was a professor at Columbia, part of the Chicago School of Economics and considered by many to be the father of modern labor economics (the Society of Labor Economics gave Mincer the organization's first-ever career achievement award in 2004 and subsequently named the award the Mincer Award).

Mincer first formulated what came to be known as the Mincer equations in his 1974 book Schooling, Experience and Earnings. The equations predicted earnings as a function of years of schooling and years of work experience using data from the 1950 and 1960 US Censuses. A 2003 paper summarized the impact of this work on the profession:

Jacob Mincer's model of earnings (1974) is a cornerstone of empirical economics. It is the framework used to estimate returns to schooling, returns to schooling quality, and to measure the impact of work experience on male-female wage gaps. It is the basis for economic studies of education in developing countries and has been estimated using data from a variety of countries and time periods. Recent studies in economic growth use the Mincer model to analyze the relationship between growth and average schooling levels across countries.

The new LAYS measure was only formally proposed and published by the World Bank in September 2018. That the simple years metric survived so long is remarkable. It is even more so that in introducing the new measure, the author's needed to spend significant time motivating for the change by justifying a claim that should be obvious to even the most naive observer of education: "schooling is not the same as learning".

Not surprisingly, they show that learning varies significantly across countries for the same amount of schooling:

In Nigeria, for example, 19 percent of young adults who have completed only primary education are able to read; by contrast, 80 percent of Tanzanians in the same category are literate. At any completed level of education, adults in some countries have learned much more than adults in other countries. (See Figure 1.1. - reproduced below)

As you might expect, these differences in learning also correlate with notable differences in people's lives. The World Bank reviewed studies that looked at the impact of learning controlled for years of schooling, which found higher learning levels are correlated with higher individual earnings, fewer deaths in childbirth, and social mobility beyond that explained by years of schooling.

Adjusting years of schooling for learning is much better than the previous metric, which completely ignored quality. But as the use in this pandemic analysis shows, it still puts most of the attention on years of schooling as the primary lever under policy makers' control. That is in part because the quality factor is calculated from comparative performance on international tests. This would be a bit like creating a quality factor to compare years of schooling in two US schools by adjusting for the average SAT scores of their students. The test scores are in part related to the quality of schooling, but the relationship between variables a teacher or school control and the test scores is a black box. As a result, the quality part of the equation is not very actionable. In contrast, the years of schooling part is directly actionable, and, in this model (and the previous reining model), more years of school equals higher earnings.

For cross-country comparisons of schooling, such a model may be appropriate. There is an incredible amount of variation in school systems within countries, so any holistic comparison is necessarily going to involve a high degree of abstraction. Moreover, getting comparable data across an extensive set of countries is difficult. The required data (years of schooling and the test score information) can be collected regularly and consistently.

But in other contexts, this model is less suitable. In particular, in addition to focusing attention on years of schooling as a major policy lever, the model embeds several strong assumptions that contribute to a clock-time mindset that is pervasive in our education system. Two particular assumptions stand out:

1. Time (adjusted for quality) is the only thing that matters. In this model, finishing college in three years would predict a lower earnings premium than finishing in the standard four.
2. All time in school is the same: In this model, only the total number of years of schooling matters. It predicts that a year of elementary school has the same impact on lifetime earnings as a year of high school. Missing a year in the middle is the same as leaving school a year earlier.

This model is not designed to predict earnings for an individual person. But on the individual level, everyday experience tells us that these simplifying assumptions obviously break down. Not every class or school year has the same impact. But, with the help of models like these, this kind of thinking is embedded in national policy thinking and, in turn, percolates to district- and school-level decision making.

This model is both a product of and a contributor to our cultural view of education that incorrectly privileges the passage of time as a sign of progress. In Dark Horse, Ogi Ogas and Todd Rose capture the absurdity of basing education on what they call "standardized time":

In the United States, most undergraduate programs require exactly four years (or 120 credit hours) to get a bachelor’s degree, whether you are majoring in marketing or marine biology or Mandarin, whether you are attending a large public land-grant university or a small private liberal arts college, whether you are taught by Nobel Prize–winning professors or distracted teaching assistants—whether you learn fast or slow.

Using the credit hour system, most colleges discourage and even erect barriers to completing bachelor's degrees in fewer than four years. In Hire Education: Mastery, Modularization and the Workforce Revolution, Michelle R. Weise and Clayton Christensen explain the credit hour's origin, adding even more to the absurdity:

Institutions of higher learning do everything possible to measure fixed seat time by relying heavily on the Carnegie unit, or the credit hour. Unfortunately, however, “the credit hour was never intended to be a measure of, or proxy for student learning.” As Amy Laitinen, deputy director of the New America Foundation, wrote, “Andrew Carnegie set out to fix a problem that had nothing to do with high school courses: the lack of pensions for college professors.” What began as a way of accessing Carnegie’s pension program quickly became the building block of every college course and degree program as well as a signaling mechanism of educational quality. Colleges and society now attribute a bachelor’s degree to the accumulation of 120 hours of course work. This numerical output has strangely become a proxy for quality even though there are no standard assessments tied to measuring this time- based learning—in other words, there is no assurance that a student has accomplished anything meaningful in this time.

In our traditional education system, everything can be captured by how much time has passed. If you ask a student how far they are in their degree, they will likely respond without hesitation using a reference to time: "I'm a junior" or "I have one year left."

This clock-time mindset pervades primary and secondary education, too, in myriad ways, for instance:

• In normal school years, schools are subject to strict laws on how many days must be included in a school year and how many hours must be in a school day but not similarly strict laws on outcomes or quality.
• For students in primary and secondary school, attendance is treated much more seriously than performance.

Time is not irrelevant. Research on memory shows that practicing recalling something over time is essential to long-term memory formation. Socialization also depends in part on the passage of time. But the current system dramatically overemphasizes the importance of time. That is a great mistake.

This same time-based mindset often also carries over to the workplace: seniority and promotions often depend on years of experience in a given role. Job postings focus on years of experience as though all years are the same or close enough. In hiring decisions, firms or individual hiring managers may use their own heuristics to make adjustments: perhaps two years as an engineer at Facebook is considered roughly equivalent to three years at a less well-known company. But time is one of the most prominent requirements and usually the only quantified ones for employees.

These rough assumptions may be OK for making broad international comparisons over time, but they provide terrible guidance for our own personal education decisions. Unfortunately, our education system and schools are built more or less around these assumptions being true.

But what if these assumptions aren't correct? Then this clock-time view of education would be misguided.

I expect that just making these assumptions explicit will make you suspicious of them. We intuitively know that things can be learned faster or slower. We know that it wouldn't make much sense to expect lower earnings for someone who finished college in three years instead of four.

The 2018 paper establishing the quality adjustment includes evidence that the focus on years of schooling may be misplaced and that adjusting for quality does not go far enough. They cite a study on the relationship between schooling and economic growth that found that what mattered was students' abilities as measured by test scores only not years of schooling and test scores:

While the relationship between test scores and growth is strong even after controlling for the years of schooling completed, years of schooling do not predict growth once test scores are taken into account, or they become only marginally significant” (WDR 2018, citing Hanushek and Woessmann 2012; see Figure 1.2 - reproduced below).

The chart on the left of Figure 1.2 shows a strong relationship between test scores and GDP per capita when controlling for years of schooling. The chart on the right shows that the relationship between years of schooling and GDP per capita is not statistically significant when controlling for test scores.

The assumption that all years are the same is equally suspect. The 2018 paper authors come to the conclusion that "learning trajectories have a plausibly local linear trend, especially across the grades that we are interested in." By grades they are interested in, they restrict their focus to grades 6-10. To reach this conclusion, they go through a tortured discussion of data that clearly shows an S-curve, non-linear trend over a wider range of grades.

For students and workers, repeated interaction with a time-based education system encourages a rate-limiting mindset. There is no incentive to learn the material faster--you can't accelerate the course, and you can't accelerate your overall degree completion. Moreover, as long as a student gets through the school year and passes their classes (which may or may not require competence in or mastery of the subject), they will move on regardless. So what's the point of trying too hard? Without any incentive to try to learn more quickly and efficiently, students don't even learn what their potential is to do so. This is a bit like teaching someone to drive a manual transmission car but not showing them that they can shift out of first gear.

Some education innovators are taking a different approach. In the last decade, a number of colleges and universities, including Southern New Hampshire University and the University of Wisconsin have launched competency-based education (CBE) degree programs. CBE programs take advantage of the increased flexibility and personalization capabilities of modern online learning platforms. Students can proceed through the material at their own pace, with completion of a course based on proven mastery of the material.

Innovation in this direction by other players has been limited: traditional colleges and universities are organized around clock time and the credit hour, and they don't show signs of changing. But there is no reason for parents, students, teachers or school administrators to accept a simple clock time mindset for education as they navigate the pandemic and beyond.

A crisis like the current pandemic will cause many direct and indirect changes to how we live and work. It may also cause us to see our former circumstances with a new perspective. Perhaps navigating the shock to our system caused by the pandemic will help us see our reliance on clock time in a new light. We could all benefit by switching to a different way of conceptualizing progress in education.

Hi, Patrick here! Welcome to this week's newsletter from Remixing Work.

Remixing Work is for the entrepreneurs, executives and educators shaping the future of work and education

First, thank you to everyone who shared the last edition. Also, a special welcome to our new readers. We've had a 25% increase in subscribers since then.

If someone forwarded this to you, you can subscribe here.

Second, I appreciated the comments I received from some of you by mail. Starting in this edition, you'll also find one-click feedback buttons at the bottom of each newsletter. It will be helpful to hear what you think so I can adjust the newsletter going forward, so please use them. Thanks!

$10 Trillion & Time: Schooling, Learning & Earning

The Northern Hemisphere school year is beginning, yet there is no end in sight to the COVID-19 pandemic. The short-term impact on students, teachers and parents is substantial and quite visible. The long-term impact on students and society will be significant but is quite uncertain.

To bring attention to the potential long-term effects and to build the case for taking additional mitigation actions, a group of World Bank economists estimated the potential long-term impact of the COVID-19 pandemic quantitatively. They published their analysis in a report in May but have been promoting the findings more recently. In a summary posted on Brookings at the end of July they write:

The students currently in school stand to lose $10 trillion in labor earnings over their work life. To get a sense of the magnitude, this sum is one-tenth of global GDP, or half of the annual economic output of the United States, or twice the global annual public expenditure on primary and secondary education.

The report's framing around their $10-trillion estimate is attention-grabbing but mostly unhelpful--it's not tied to actionable policy options and the uncertainty involved is huge and not sufficiently characterized. Beyond the top-line number, the report identifies several negative short-term and long-term effects the pandemic will have on students around the world.

In applying the standard models for assessing the returns of education to the shock of the pandemic, the report also casts some of the major assumptions of such models in a new light. These assumptions are deeply embedded in policy thinking about education. They likely also appear in your own views about education at times. But these assumptions are often inappropriate and can be quite limiting. I will share a bit more about the report and then return to this point below.

The report identifies three main impacts of school closures:

• Increased dropout rates
• Missed learning for the time school is closed
• Increased learning loss from the longer period out of school

The increased dropout rates stand out as the most concerning and non-obvious effect because students who drop out miss the schooling they would have received during the closures (like all students) and miss the subsequent period they would have remained in school if there wasn't the disruption.

The other two effects are notable, but they also bring to light deeply ingrained assumptions about how education translates to learning and how learning translates to higher earnings.

The report's main model is built around a metric called learning-adjusted years of schooling (LAYS). LAYS is a country-level measure of the number of years of schooling that students in that country complete on average, adjusted by a quality factor for that country. To estimate the impact of school disruptions, the report calculates the pandemic's effect on LAYS. It then translates that into a change in earnings using returns to education estimated elsewhere.

The LAYS  metric was developed and published in 2018 by several other World Bank economists to replace the metric previously used for cross-country comparisons, which only considered years of schooling without any quality adjustment. Except in its use of the s newer quality-adjusted metric, the report's approach using years of schooling as the primary input variable has been the standard methodology in labor economics for nearly 50 years.

The origin of using years of schooling to explain variation in earnings was the ground-breaking empirical work of Jacob Mincer. Mincer was a professor at Columbia, part of the Chicago School of Economics and considered by many to be the father of modern labor economics (the Society of Labor Economics gave Mincer the organization's first-ever career achievement award in 2004 and subsequently named the award the Mincer Award).

Mincer first formulated what came to be known as the Mincer equations in his 1974 book Schooling, Experience and Earnings. The equations predicted earnings as a function of years of schooling and years of work experience using data from the 1950 and 1960 US Censuses. A 2003 paper summarized the impact of this work on the profession:

Jacob Mincer's model of earnings (1974) is a cornerstone of empirical economics. It is the framework used to estimate returns to schooling, returns to schooling quality, and to measure the impact of work experience on male-female wage gaps. It is the basis for economic studies of education in developing countries and has been estimated using data from a variety of countries and time periods. Recent studies in economic growth use the Mincer model to analyze the relationship between growth and average schooling levels across countries.

The new LAYS measure was only formally proposed and published by the World Bank in September 2018. That the simple years metric survived so long is remarkable. It is even more so that in introducing the new measure, the author's needed to spend significant time motivating for the change by justifying a claim that should be obvious to even the most naive observer of education: "schooling is not the same as learning".

Not surprisingly, they show that learning varies significantly across countries for the same amount of schooling:

In Nigeria, for example, 19 percent of young adults who have completed only primary education are able to read; by contrast, 80 percent of Tanzanians in the same category are literate. At any completed level of education, adults in some countries have learned much more than adults in other countries. (See Figure 1.1. - reproduced below)

As you might expect, these differences in learning also correlate with notable differences in people's lives. The World Bank reviewed studies that looked at the impact of learning controlled for years of schooling, which found higher learning levels are correlated with higher individual earnings, fewer deaths in childbirth, and social mobility beyond that explained by years of schooling.

Adjusting years of schooling for learning is much better than the previous metric, which completely ignored quality. But as the use in this pandemic analysis shows, it still puts most of the attention on years of schooling as the primary lever under policy makers' control. That is in part because the quality factor is calculated from comparative performance on international tests. This would be a bit like creating a quality factor to compare years of schooling in two US schools by adjusting for the average SAT scores of their students. The test scores are in part related to the quality of schooling, but the relationship between variables a teacher or school control and the test scores is a black box. As a result, the quality part of the equation is not very actionable. In contrast, the years of schooling part is directly actionable, and, in this model (and the previous reining model), more years of school equals higher earnings.

For cross-country comparisons of schooling, such a model may be appropriate. There is an incredible amount of variation in school systems within countries, so any holistic comparison is necessarily going to involve a high degree of abstraction. Moreover, getting comparable data across an extensive set of countries is difficult. The required data (years of schooling and the test score information) can be collected regularly and consistently.

But in other contexts, this model is less suitable. In particular, in addition to focusing attention on years of schooling as a major policy lever, the model embeds several strong assumptions that contribute to a clock-time mindset that is pervasive in our education system. Two particular assumptions stand out:

1. Time (adjusted for quality) is the only thing that matters. In this model, finishing college in three years would predict a lower earnings premium than finishing in the standard four.
2. All time in school is the same: In this model, only the total number of years of schooling matters. It predicts that a year of elementary school has the same impact on lifetime earnings as a year of high school. Missing a year in the middle is the same as leaving school a year earlier.

This model is not designed to predict earnings for an individual person. But on the individual level, everyday experience tells us that these simplifying assumptions obviously break down. Not every class or school year has the same impact. But, with the help of models like these, this kind of thinking is embedded in national policy thinking and, in turn, percolates to district- and school-level decision making.

This model is both a product of and a contributor to our cultural view of education that incorrectly privileges the passage of time as a sign of progress. In Dark Horse, Ogi Ogas and Todd Rose capture the absurdity of basing education on what they call "standardized time":

In the United States, most undergraduate programs require exactly four years (or 120 credit hours) to get a bachelor’s degree, whether you are majoring in marketing or marine biology or Mandarin, whether you are attending a large public land-grant university or a small private liberal arts college, whether you are taught by Nobel Prize–winning professors or distracted teaching assistants—whether you learn fast or slow.

Using the credit hour system, most colleges discourage and even erect barriers to completing bachelor's degrees in fewer than four years. In Hire Education: Mastery, Modularization and the Workforce Revolution, Michelle R. Weise and Clayton Christensen explain the credit hour's origin, adding even more to the absurdity:

Institutions of higher learning do everything possible to measure fixed seat time by relying heavily on the Carnegie unit, or the credit hour. Unfortunately, however, “the credit hour was never intended to be a measure of, or proxy for student learning.” As Amy Laitinen, deputy director of the New America Foundation, wrote, “Andrew Carnegie set out to fix a problem that had nothing to do with high school courses: the lack of pensions for college professors.” What began as a way of accessing Carnegie’s pension program quickly became the building block of every college course and degree program as well as a signaling mechanism of educational quality. Colleges and society now attribute a bachelor’s degree to the accumulation of 120 hours of course work. This numerical output has strangely become a proxy for quality even though there are no standard assessments tied to measuring this time- based learning—in other words, there is no assurance that a student has accomplished anything meaningful in this time.

In our traditional education system, everything can be captured by how much time has passed. If you ask a student how far they are in their degree, they will likely respond without hesitation using a reference to time: "I'm a junior" or "I have one year left."

This clock-time mindset pervades primary and secondary education, too, in myriad ways, for instance:

• In normal school years, schools are subject to strict laws on how many days must be included in a school year and how many hours must be in a school day but not similarly strict laws on outcomes or quality.
• For students in primary and secondary school, attendance is treated much more seriously than performance.

Time is not irrelevant. Research on memory shows that practicing recalling something over time is essential to long-term memory formation. Socialization also depends in part on the passage of time. But the current system dramatically overemphasizes the importance of time. That is a great mistake.

This same time-based mindset often also carries over to the workplace: seniority and promotions often depend on years of experience in a given role. Job postings focus on years of experience as though all years are the same or close enough. In hiring decisions, firms or individual hiring managers may use their own heuristics to make adjustments: perhaps two years as an engineer at Facebook is considered roughly equivalent to three years at a less well-known company. But time is one of the most prominent requirements and usually the only quantified ones for employees.

These rough assumptions may be OK for making broad international comparisons over time, but they provide terrible guidance for our own personal education decisions. Unfortunately, our education system and schools are built more or less around these assumptions being true.

But what if these assumptions aren't correct? Then this clock-time view of education would be misguided.

I expect that just making these assumptions explicit will make you suspicious of them. We intuitively know that things can be learned faster or slower. We know that it wouldn't make much sense to expect lower earnings for someone who finished college in three years instead of four.

The 2018 paper establishing the quality adjustment includes evidence that the focus on years of schooling may be misplaced and that adjusting for quality does not go far enough. They cite a study on the relationship between schooling and economic growth that found that what mattered was students' abilities as measured by test scores only not years of schooling and test scores:

While the relationship between test scores and growth is strong even after controlling for the years of schooling completed, years of schooling do not predict growth once test scores are taken into account, or they become only marginally significant” (WDR 2018, citing Hanushek and Woessmann 2012; see Figure 1.2 - reproduced below).

The chart on the left of Figure 1.2 shows a strong relationship between test scores and GDP per capita when controlling for years of schooling. The chart on the right shows that the relationship between years of schooling and GDP per capita is not statistically significant when controlling for test scores.

The assumption that all years are the same is equally suspect. The 2018 paper authors come to the conclusion that "learning trajectories have a plausibly local linear trend, especially across the grades that we are interested in." By grades they are interested in, they restrict their focus to grades 6-10. To reach this conclusion, they go through a tortured discussion of data that clearly shows an S-curve, non-linear trend over a wider range of grades.

For students and workers, repeated interaction with a time-based education system encourages a rate-limiting mindset. There is no incentive to learn the material faster--you can't accelerate the course, and you can't accelerate your overall degree completion. Moreover, as long as a student gets through the school year and passes their classes (which may or may not require competence in or mastery of the subject), they will move on regardless. So what's the point of trying too hard? Without any incentive to try to learn more quickly and efficiently, students don't even learn what their potential is to do so. This is a bit like teaching someone to drive a manual transmission car but not showing them that they can shift out of first gear.

Some education innovators are taking a different approach. In the last decade, a number of colleges and universities, including Southern New Hampshire University and the University of Wisconsin have launched competency-based education (CBE) degree programs. CBE programs take advantage of the increased flexibility and personalization capabilities of modern online learning platforms. Students can proceed through the material at their own pace, with completion of a course based on proven mastery of the material.

Innovation in this direction by other players has been limited: traditional colleges and universities are organized around clock time and the credit hour, and they don't show signs of changing. But there is no reason for parents, students, teachers or school administrators to accept a simple clock time mindset for education as they navigate the pandemic and beyond.

A crisis like the current pandemic will cause many direct and indirect changes to how we live and work. It may also cause us to see our former circumstances with a new perspective. Perhaps navigating the shock to our system caused by the pandemic will help us see our reliance on clock time in a new light. We could all benefit by switching to a different way of conceptualizing progress in education.

Further reading:

The Transcend Newsletter this week also looked at the World Bank's COVID-19 report.

• Using the lens of the United Nations' Sustainable Development Goals, it emphasizes the pandemic's impact on students in middle- and low-income countries and looks at how school systems in those countries are adapting (including using radio and SMS-based solutions for remote learning).

What do you think about this edition?

Amazing | Good | Meh | Bad

August 7, 2020

Hi all,​

This week's post is a longer, more inspirational piece than usual. I hope you enjoy it.

I love when I receive comments from readers, so feel free to share any thoughts or reactions--feedback is a gift. :)

And sharing is an even better gift. So if you know someone else who might enjoy reading it, please pass it on.

If someone shared this with you, you can subscribe here.​

Patrick

Conceptualizing our potential

In 1965, Intel's co-founder Gordon Moore predicted that the number of components in an integrated circuit would double every year until 1975. The forecast proved remarkably accurate. In 1975, he extended his prediction at a slower rate of two years per doubling. His prescient prediction continued to hold correct until recently.

This astounding and regular progress has been one of the driving forces of the seemingly inevitable improvements in computer power and affordability that we have experienced in the past few decades. The chart below shows the increasing speed of processors over time on a log scale.

Moore based his forecast on the brief historical trends in the nascent semiconductor industry and his understanding of the manufacturing process. But in making his prediction, there was nothing preordained that the "law" should persist.

David Rotman recently recounted:

Moore wrote that "cramming more components onto integrated circuits," the title of his 1965 article, would "lead to such wonders as home computers—or at least terminals connected to a central computer—automatic controls for automobiles, and personal portable communications equipment." In other words, stick to his road map of squeezing ever more transistors onto chips and it would lead you to the promised land. And for the following decades, a booming industry, the government, and armies of academic and industrial researchers poured money and time into upholding Moore's Law, creating a self-fulfilling prophecy that kept progress on track with uncanny accuracy. Though the pace of progress has slipped in recent years, the most advanced chips today have nearly 50 billion transistors.

The prophesy's fulfillment became expected, and that expectation became destiny, in part, "because the semiconductor industry decided it would," Rotman writes. The law gave the industry a clear R&D target and a drumbeat-like cadence. Much like Apple has for years organized all of its iPhone hardware and software development around an annual product release cycle, Moore's law kept continuous pressure on semiconductor manufacturers to advance the state of the art.

A similarly steady beat has marked progress in extending human life expectancy. Until a slight reversal after 2014, life expectancy in the US has increased an average of two years with each passing decade. Unlike for Gordon Moore's exponential growth, this growth is linear—but it has still been steady.

There is no famous "law" that has served as a drumbeat for pursuing increasing longevity. Nevertheless, these extra years of expected life are undeniable evidence of our potential to improve ourselves—against what many consider a rather important dimension of human performance.

Improving other dimensions of human performance can happen at the level of a single person and can occur much faster. Athletics provides many such examples of transcending what previously seemed to be boundaries of human potential.

In 1954, Roger Bannister became the first person to run a mile in less than four minutes. It was a goal long pursued but long out of reach. Bill Taylor described how substantial the barrier had seemed:

[R]unners had been chasing the goal seriously since at least 1886, and that the challenge involved the most brilliant coaches and gifted athletes in North America, Europe, and Australia. "For years, milers had been striving against the clock, but the elusive four minutes had always beaten them," he notes. "It had become as much a psychological barrier as a physical one. And like an unconquerable mountain, the closer it was approached, the more daunting it seemed."

Having stood as a barrier for decades, many thought Bannister would be unique in that accomplishment for some time. Instead, he quickly had company. Only 46 days later, John Landy broke the four-minute mark and Bannister's record time. A year later, three runners finished the same race in under four minutes. Over the last 50 years, more than a thousand runners have accomplished a feat that previously may have seemed superhuman (to use the contemporary phrase).

The fastest have gotten even faster still. The world record is now held by Hicham El Guerrouj, with a time of 3 minutes, 43 seconds and 13 hundredths.

These athletic feats are a powerful illustration of the potential to improve even our most basic physical capabilities. What if we could similarly enhance our cognitive capabilities?

Almost no one's livelihood depends on how fast they run. But virtually everyone's livelihood depends on how well they think. If we could each double our cognitive capabilities, we would be individually and collectively much better off.

The obvious objection is that this isn't possible. But there is good evidence that we can improve at least some of our cognitive abilities on scales much greater than a mere doubling.

For instance, there is substantial evidence that dramatic improvements in memory are possible for anyone committed to training. Journalist Joshua Foer demonstrated what is possible in his account of his memory training journey in his 2011 book "Moonwalking with Einstein."

Reigning world memory champion Ben Pridmore's ability to memorize "the precise order of 1,528 random digits in an hour" and "any poem handed to him" sparked Foer's curiosity.  It burst into flames when he read Pridmore's explanation:

"It's all about technique and understanding how the memory works," he told the reporter. "Anyone could do it, really."

Alongside investigating the people who compete in memory tournaments, Foer decided to compete too. After a year of intense practice, Foer won the US Memory Championship. Along with completing various other feats, Foer memorized every card in a thoroughly shuffled deck faster than all of his competitors.

The techniques Foer learned to memorize random strings of numbers aren't the solution to learning the things that matter to us in school and work. But this narrow ability is still remarkable given humans usually can only remember "seven [items], plus or minus two." That he and others can remember lengths of higher orders of magnitude suggests the correct scale to conceptualize our cognitive abilities is much greater than we commonly think.

Consider this in the abstract: If you score 90 on some measure of performance and believe the maximum score possible is 100, then you don't have too much room for improvement. Maxing out the scale would only make you ~10% better. If it seemed hard or risky to improve, it probably wouldn't be worth it: you don't have much to gain and have plenty you could lose.

But what if your belief about the maximum score is entirely wrong? What if the maximum is actually 1,000? Then there are tremendous potential gains ahead of you.

How we conceive of our potential has significant effects. If you thought taking care of your health might give you another hundred years of life, wouldn't you act differently than if you thought it might buy you an extra decade at most?

Unfortunately, it is common to think of our cognitive abilities as if the scale only goes to 100. This view is partly related to our education system's performance, which has not generated gains that look like Moore's law or our increasing longevity. The cause could be a lack of understood and accepted measures for our cognitive abilities. Perhaps thousands of people have learned to think at a four-minute pace.

But there is no readily available data to support that narrative. Most of the available data related in some way to aggregate cognitive performance is from our education system, where the indicators are quite mixed. I won't draw any conclusions about our education system in this essay—there are both successes and failures that could be analyzed. But, notably, there is no easily accessible story of increasing capability. In inputs and hours spent, yes. But in outputs that matter, it's not so clear. Dysfunctions burden our healthcare system, but healthcare can at least tell a story of progress based on our increasing longevity.

We devote enormous resources to education, and many dedicated people have worked tirelessly to improve our education system. Yet, with perceived results failing to match the perceived effort, many people doubt that future initiatives to improve education will have much impact.

This narrative in which substantial effort translates into minimal results is even more destructive: it shrinks our perceived potential. It suggests the scale really only goes to 100.

But the performance of the memory champions suggests this is misguided. With the right techniques, they cannot only memorize a few more cards than the average person--they can memorize a few more decks of cards. It suggests that substantially more powerful cognitive abilities may be available to all of us.

Unfortunately, stories of great human achievement are often told in a way that inhibits other people rather than expanding their sense of potential. Instead of holding up people with notable accomplishments in a way that says "Look, anyone can walk this path if they want," cultural storytellers instead often spin a tale of "superhuman" ability based on prodigal "genius" or "talent." Potential progress needs to seem accessible. We only get better by being challenged, but most people will give up if the challenge seems too insurmountable.

"Superhuman" sounds impossible. Using it to describe the impressive abilities of people that are all too human is misuse. More often than not, those abilities are the product of years of deliberate practice, which is accessible to all.

There is much we don't know about how learning and intelligence work and how to optimize them. Yet the existing knowledge is tremendous compared to how much we actually apply. We could all be better learners and more cognitively capable with deliberate practice and different approaches to learning. This will require changes to how we learn at school and work and changes to the institutions in which we learn. These changes are easy to dismiss if we don't appreciate our potential.

The scale doesn't stop at 100. Our potential is greater than we realize.

A permanent version of this post can be found here.

Hi there,

I hope you've been managing through the pandemic OK and that you and your family are well.

After a hiatus, Remixing Week is back to a weekly schedule. The schedule will shift to mid-week starting next week onwards.

This week looks at the interesection between the future of work and the future of (higher) education. They are inseparable in many ways, and I'll be analyzing that relevance in a number of upcoming weeks.

On to the Work Mix....

A/ From Higher Ed to Hire Ed

The last decade's data and discourse reveal an unrealized American Dream for many. There are many reasons for this disappointment, and the education system's role is significant. Individual advancement and societal advancement depend on learning, skill acquisition and pathways to good jobs, and our higher education system is falling short.

Our education system doesn't serve a single purpose: But preparing people to live productive lives is undoubtedly a vital part. At a minimum, that preparation should include obtaining the skills and capabilities needed to successfully enter the workforce in a good job and to be capable of further development after that.

Higher education providers aren't direct players in the labor market—workers and employers are—but they play an essential role. In the past, employers took on more responsibility for training and developing employees and provided opportunities for internal advancement. The design of most colleges and universities did not aim to directly prepare students for jobs—unless those were jobs in the academy (and only after further study).

But the labor market has changed for good. In today's economy, employers expect new employees to already have more of the skills required for a job, and employees are also more responsible for navigating their career advancement. These effects are widely lamented, but there are no serious policy proposals that would reverse them. In the face of these changing conditions, our higher education system has failed to adapt significantly.

Americans have been slowly becoming more aware of the problems with our current higher ed model, chief among them terrible completion rates and rising costs. Now the current pandemic—and the prospect of paying $50,000 for a year of lectures on Zoom—has accelerated the arrival of a collective "aha!" moment.

In the rest of the post I'll address:

• Why higher ed should focus more on employability and setting students up for successful careers
• Three long-term changes in the labor market that the education system needs to respond to (but hasn't)
• The implications for existing colleges and universities and for new higher ed providers

​

Higher education providers should prioritize their students' employability and career success

This point seems incredibly obvious to most people but not to all higher education programs, so I'll highlight some key points:

• Students want jobs and see this as a primary objective in pursuing a four-year degree.
• Good jobs lead to higher salaries. Of course, money isn't everything, but below a certain level of earning, it has a massive impact on people's health and quality of life
• Moreover, work is an important source of meaning for many people. The autonomy, flexibility and security that come with better, higher-paying jobs make it much easier for people to focus on pursuing that meaning rather than just trying to survive.
• Better jobs also come with other benefits like higher status and less risk of various kinds.

Despite this, most colleges and universities do not prioritize employability and career success for their students. This is reflected in the metrics they track and report on, their allocation of resources, their pay and incentive systems—and even in the public comments by faculty and administrators at some schools

Some defenders of the status quo say that college isn't about getting a job and that it should be a time for exploration. This is a false trade-off: the college experience can comprise many other things besides job preparation too. Given the costs involved and the inescapability of the real world waiting after graduation, everyone is better served if students have a smooth pathway to a good first job.

Other defenders of the status quo emphasize that college isn't about job skills: It's about teaching you how to learn and think. I believe that is true and that those are sought-after skills for a wide range of careers. However, colleges do not explicitly target the development of even these general cognitive skills, based on how students are evaluated and what metrics and incentives the schools use.

​

The labor market has changed

Our economy as a whole has become much more dynamic over the past century. In my last post, I shared data showing the accelerating turnover of the Fortune 500 since the mid-20th century. The largest companies do not hold their space on the rankings as long as they used to. The emergence of completely new industries has played an important part. But almost every industry has one or more players founded after the year 2000 in the Fortune 500.

In response to and in conjunction with this more dynamic economy, the labor market has also become more dynamic. The idea of jobs for life is long gone, as companies restructure and conduct layoffs swiftly when under pressure. Workers have also learned to change jobs more frequently to get ahead. Automation and outsourcing have slowly eliminated categories of work, while entirely new roles (e.g., social media marketer) have emerged. Employers have collectively shifted more risk to employees in the form of retirement and healthcare benefits and have less willingness to invest in training and development, even as the technical content of roles have increased.

More and more, employers want to hire workers who already have the required skills and who will be productive very quickly

Also, there are many more college graduates, meaning being a graduate in and of itself is less of a differentiator. This has led to the pursuit of more and more graduate degrees—an arms race as they say.

​

Implications for higher education

In the face of this new labor market, the U.S. higher education system has mostly doubled down on the same model that has been in place for more than a hundred years. It is not working.

​

More focus is needed in three areas

(1) Increase focus on navigation and getting real-world feedback to gather experience

• Companies don't make significant investments without running small experiments to test their hypotheses first. After they decide to proceed with an investment, they still review it at key milestones with pre-determined criteria to decide whether to proceed further or change course.
• In contrast, the traditional college model does not enable real-world experimentation (it allows some experimentation among fields of study but those do not provide much information about what different jobs are like). Moreover, the traditional model encourages completion regardless of any new data.

​

(2) Increase focus on the skills employers want

• First jobs are important for all the reasons discussed above. Providers of higher ed should explicitly seek to develop relevant job skills—including relevant process and technical skills.
• Critical thinking skills are incredibly valuable but programs should be explicitly designed to develop these.

​

(3) Increase focus on connecting students with employers

• Career services are woeful at most colleges

​

To address these areas and to respond to a more dynamic labor market, shorter programs are needed. For adults in their 20s, 30s or 40s looking to change careers, this is obvious. But shorter programs are also needed for new workers.

Shorter programs would allow students to get real-world experience sooner and then return for more targeted and more advanced education, much like MBA students who are expected to have at least 2-4 years of work experience before attending. Traditional bachelor's degrees are too long for this application, and the right kind of complementary follow-up programs don't exist. Four-year degrees may still make sense for some students but perhaps fewer than you think.

More new institutions are needed. In theory, existing higher ed institutions could adapt to meet the needs of current students. In practice, this is not happening. To be clear, this is not to say that the current institutions should not exist in the future and can't be part of the solution. It's just to say that new institutions will be required to deliver the necessary degree of innovation. As I noted in the previous post, the average founding year of the top 50 national universities (according to U.S. News) is 1846. None in the top 50 were founded in the last 50 years.

In comparison to the changes in our broader economy, the innovation by traditional colleges and universities has been quite minimal. But there is still time to change, and there is more of an imperative than ever. The first wave of online education providers like Coursera and Udacity gained a lot of attention but haven't had a significant impact on the core system. There is also a new wave of startups taking on this space in exciting ways.

Education is key to the future of work. In the next weeks and months, I'll be covering these topics in more detail, including:

• Why is the status quo so sticky? Why are most colleges so old? Why isn't there more innovation?
• How the current model is driving high costs and student debt. What does this mean for students
• How has access to higher education improved and where are the largest gaps
• The landscape of newer higher education providers, especially "last mile" programs—including boot camps, apprenticeship and college alternatives—that are putting employability first
• Why for-profit colleges have mostly been a failure and the implications for education startups
• What other elements of today's colleges could and should be unbundled
• How elite universities could lead change from the top

​

B/ One quick listen

This podcast on the Strada Education Network features Oren Cass, author of The Once and Future Worker.

​https://stradaeducationnetworklessonsearned.libsyn.com/episode-2-oren-cass-0​

He says we’ve gone too far in our college-going culture, steering too many students down a postsecondary pathway that only serves about one-third of our population.

I will be exploring a broader set of ideas in this space in the coming weeks but he does a good job calling out many of the assumptions we have built into our system.

Our education system doesn't serve a single purpose: But preparing people to live productive lives is undoubtedly a vital part. At a minimum, that preparation should include obtaining the skills and capabilities needed to successfully enter the workforce in a good job and to be capable of further development after that.

Higher education providers aren't direct players in the labor market—workers and employers are—but they play an essential role. In the past, employers took on more responsibility for training and developing employees and provided opportunities for internal advancement. The design of most colleges and universities did not aim to directly prepare students for jobs—unless those were jobs in the academy (and only after further study).

But the labor market has changed for good. In today's economy, employers expect new employees to already have more of the skills required for a job, and employees are also more responsible for navigating their career advancement. These effects are widely lamented, but there are no serious policy proposals that would reverse them. In the face of these changing conditions, our higher education system has failed to adapt significantly.

Americans have been slowly becoming more aware of the problems with our current higher ed model, chief among them terrible completion rates and rising costs. Now the current pandemic—and the prospect of paying $50,000 for a year of lectures on Zoom—has accelerated the arrival of a collective "aha!" moment.

In the rest of the post I'll address:

• Why higher ed should focus more on employability and setting students up for successful careers
• Three long-term changes in the labor market that the education system needs to respond to (but hasn't)
• The implications for existing colleges and universities and for new higher ed providers

Higher education providers should prioritize their students' employability and career success

This point seems incredibly obvious to most people but not to all higher education programs, so I'll highlight some key points:

• Students want jobs and see this as a primary objective in pursuing a four-year degree.
• Good jobs lead to higher salaries. Of course, money isn't everything, but below a certain level of earning, it has a massive impact on people's health and quality of life
• Moreover, work is an important source of meaning for many people. The autonomy, flexibility and security that come with better, higher-paying jobs make it much easier for people to focus on pursuing that meaning rather than just trying to survive.
• Better jobs also come with other benefits like higher status and less risk of various kinds.

Despite this, most colleges and universities do not prioritize employability and career success for their students. This is reflected in the metrics they track and report on, their allocation of resources, their pay and incentive systems—and even in the public comments by faculty and administrators at some schools

Some defenders of the status quo say that college isn't about getting a job and that it should be a time for exploration. This is a false trade-off: the college experience can comprise many other things besides job preparation too. Given the costs involved and the inescapability of the real world waiting after graduation, everyone is better served if students have a smooth pathway to a good first job.

Other defenders of the status quo emphasize that college isn't about job skills: It's about teaching you how to learn and think. I believe that is true and that those are sought-after skills for a wide range of careers. However, colleges do not explicitly target the development of even these general cognitive skills, based on how students are evaluated and what metrics and incentives the schools use.

The labor market has changed

Our economy as a whole has become much more dynamic over the past century. In my last post, I shared data showing the accelerating turnover of the Fortune 500 since the mid-20th century. The largest companies do not hold their space on the rankings as long as they used to. The emergence of completely new industries has played an important part. But almost every industry has one or more players founded after the year 2000 in the Fortune 500.

In response to and in conjunction with this more dynamic economy, the labor market has also become more dynamic. The idea of jobs for life is long gone, as companies restructure and conduct layoffs swiftly when under pressure. Workers have also learned to change jobs more frequently to get ahead. Automation and outsourcing have slowly eliminated categories of work, while entirely new roles (e.g., social media marketer) have emerged. Employers have collectively shifted more risk to employees in the form of retirement and healthcare benefits and have less willingness to invest in training and development, even as the technical content of roles have increased.

More and more, employers want to hire workers who already have the required skills and who will be productive very quickly

Also, there are many more college graduates, meaning being a graduate in and of itself is less of a differentiator. This has led to the pursuit of more and more graduate degrees—an arms race as they say.

Implications for higher education

In the face of this new labor market, the U.S. higher education system has mostly doubled down on the same model that has been in place for more than a hundred years. It is not working.

More focus is needed in three areas

(1) Increase focus on navigation and getting real-world feedback to gather experience

• Companies don't make significant investments without running small experiments to test their hypotheses first. After they decide to proceed with an investment, they still review it at key milestones with pre-determined criteria to decide whether to proceed further or change course.
• In contrast, the traditional college model does not enable real-world experimentation (it allows some experimentation among fields of study but those do not provide much information about what different jobs are like). Moreover, the traditional model encourages completion regardless of any new data.

(2) Increase focus on the skills employers want

• First jobs are important for all the reasons discussed above. Providers of higher ed should explicitly seek to develop relevant job skills—including relevant process and technical skills.
• Critical thinking skills are incredibly valuable but programs should be explicitly designed to develop these.

(3) Increase focus on connecting students with employers

• Career services are woeful at most colleges

To address these areas and to respond to a more dynamic labor market, shorter programs are needed. For adults in their 20s, 30s or 40s looking to change careers, this is obvious. But shorter programs are also needed for new workers.

Shorter programs would allow students to get real-world experience sooner and then return for more targeted and more advanced education, much like MBA students who are expected to have at least 2-4 years of work experience before attending. Traditional bachelor's degrees are too long for this application, and the right kind of complementary follow-up programs don't exist. Four-year degrees may still make sense for some students but perhaps fewer than you think.

More new institutions are needed. In theory, existing higher ed institutions could adapt to meet the needs of current students. In practice, this is not happening. To be clear, this is not to say that the current institutions should not exist in the future and can't be part of the solution. It's just to say that new institutions will be required to deliver the necessary degree of innovation. As I noted in the previous post, the average founding year of the top 50 national universities (according to U.S. News) is 1846. None in the top 50 were founded in the last 50 years.

In comparison to the changes in our broader economy, the innovation by traditional colleges and universities has been quite minimal. But there is still time to change, and there is more of an imperative than ever. The first wave of online education providers like Coursera and Udacity gained a lot of attention but haven't had a significant impact on the core system. There is also a new wave of startups taking on this space in exciting ways.

Education is key to the future of work. In the next weeks and months, I'll be covering these topics in more detail, including:

• Why is the status quo so sticky? Why are most colleges so old? Why isn't there more innovation?
• How the current model is driving high costs and student debt. What does this mean for students
• How has access to higher education improved and where are the largest gaps
• The landscape of newer higher education providers, especially "last mile" programs—including boot camps, apprenticeship and college alternatives—that are putting employability first
• Why for-profit colleges have mostly been a failure and the implications for education startups
• What other elements of today's colleges could and should be unbundled
• How elite universities could lead change from the top

A historical look at the Fortune 500 illustrates this increasing dynamism quantitatively:

Source: Kaufmann, https://www.kauffman.org/wp-content/uploads/2012/06/fortune_500_turnover.pdf

As the chart above shows, half of the companies included on the Fortune 500 in 1955 remained on the list 30 years later. But the 1995 list turned over twice as fast— only 14 years later fewer than half of the 1995 list remained.

The chart below provides another lens on the economy's dynamism. It arranges each of the current members of the Fortune 500 by their industry and founding year. It also indicates the mean founding year for each industry. The average founding year for the oldest industry, Household Products, is 1898. While the average founding year for the youngest industry, Telecom & Tech is 1972.

Source: Fortune, https://interactives.fortune.com/age_f500/index.html

Compared to this, colleges and universities are mostly antiques. I calculated the average founding year for U.S. News's 2019 lists of top colleges and universities. (I used U.S. News' lists for their widespread use & familiarity, not at all to endorse the lists.)

Among the top 100 liberal arts colleges:

• The mean founding year is 1856
• 11 were founded in the last 100 years
• 2 were founded in the last 50 years

Among the top 50 national universities:

• The mean founding year is 1845
• 5 were founded in the last 50 years
• Zero(!) were founded in the last 50 years

The contrast is striking. Our top colleges and universities are ~100 years older on average than our largest companies. Even the oldest industries have proportionately more new entrants among their largest players.. Almost all the industries include major companies founded after the year 2000.

This age difference alone doesn't indicate a problem. But it does raise some obvious questions:

• Is our education system well adapted to our modern economy?
• Are these traditional leaders innovating enough or Is their dominance holding innovation back?

In the next few weeks, *Remixing* *Work* will examine these questions and others to better understand what the future of work means for the future of education.

Hi there,

I hope your new year is off to a great start! Work Mix is back to its weekly schedule for the year.

Topics this week:

• A look at the most common low-wage jobs today and implications for the labor market
• A follow-up from last newsletter looking at the implementation of California's new labor law targeting gig workers

​

A/ Labor regulation follow-up as California rolls out AB5

​Last Work Mix discussed California's new labor regulations targeting gig workers. Since then, the new law AB 5 took effect on January 1, leading to a number of stories.

Uber, which perhaps was the biggest targets of the law, has made changes to its business in California that it thinks will allow it to continue classifying drivers as independent contractors despite the law's tough criteria. These changes include showing drivers the destination before they accept a ride and eliminating upfront pricing—in effect returning to its earlier model where the fare is only determined at the end of the ride based on the actual route and conditions. Uber is also rolling back some of its rewards program benefits in California related to pricing, discounts and cancellations.

Drivers are largely happy with these changes.

For riders, it's a different story. Some of these changes won't have too much impact on riders. But the option for drivers to reject a rider based on their destination is likely to have a significant impact for some riders on some routes--especially at times and in places where there are few drivers. My own experience is suggestive: In the last few years, I've had more times than I can count when a driver has refused a trip across the Bay Bridge after I've already gotten in the car—one time the driver started the trip early, saw the route and drove off as I was walking to the door. We've all had drivers call us and ask what our destination was (most often, this seems to happen at the airport). Where there are many drivers with different preferences, this won't make much of a problem. But, if you're going to an area most drivers don't want to go to, you could be out of luck for a ride or in for a very long wait.

In the meantime, Uber is not relying only on this approach: it's also battling the law directly. Its multi-pronged strategy includes working with Lyft and DoorDash to promote a statewide ballot initiative in the November election to exempt their workers and also pursuing a lawsuit against the state of California in conjunction with Postmates, arguing that the law is "irrational and unconstitutional."

Beyond Uber, concerns abound in the state and beyond. Many industries and groups of workers are angling for exemptions in anticipated follow-up legislation. In other states looking at similar laws, companies and workers continue to express concern with California's approach. As one New Jersey freelancer describes (and agonizes over) legislation under consideration that is modeled on California's law:

This legislation essentially classifies every worker as an employee unless they can pass a very narrow set of criteria. While their goal is to prevent companies from misclassifying workers, they are poorly written and will prohibit many successful, independent contractors from being able to earn a living the way they choose.

As other states and the federal government potentially look at new legislation for the gig economy, they would benefit from a different approach. In a chapter of the World Economic Forum's white paper "Realizing Human Potential in the Fourth Industrial Revolution" titled "Facilitating the Transition to a New World of Work," they recommend:

Reforming employment law and regulatory classifications to better enable independent workers to benefit from new opportunities while managing associated challenges may require an agile approach that involves modifying current policies as well as testing wholly new models, and measuring and monitoring outcomes. All relevant stakeholders will need to remain open to iterative learning.

As I said last time:

Rather than address distinct features of the modern economy, AB5 tries to force the world to conform to the models of the past.

New regulations and regulatory reform are both needed. Still, the starting point needs to be a better understanding of how the labor market currently works (especially for freelancers and the gig economy), and it needs to accommodate more models for work rather than limit them.

​

B/ Labor market facts: top low-wage jobs

I recently came across the following interesting data that frame some of the challenges and opportunities facing the labor market. The chart shows the distribution of workers in the top ten most common low-wage occupations, which represents almost half of all low-wage workers in the US.

​Hi there,

This week's note is a short one covering:

• The roll-out of new labor laws targeting the gig economy, especially the early impacts of California's, which taks effect on January 1, 2020
• The declining importance of higher education according to Americans

If you are celebrating Christmas or Hannaukuh in the coming week, I hope you enjoy!

As always, I'd love to hear any reactions or thoughts from you. If you know someone interested in these topics, thanks for passing this note on to them!

A/ New "pro-worker" labor regulations causing some damage

​California law AB5 will take effect in January, and some workers are losing their freelance positions as a result. For instance, Vox Media cut hundreds of its freelancers in the state. Its supporters passed the law to tighten labor regulations for the gig economy and limit businesses' ability to classify workers as independent contractors. But its broad drafting has many workers and firms understandably concerned. Translators, truck drivers and others have begun to find out they may be losing their freelance positions.

Uber, on the other hand, is trying to thread a needle through the requirements of the new law so that it can continue classifying workers as independent contractors. Drivers in California will now be able to see a rider's destination before accepting a ride and will be able to decline as many rides as they want without losing rewards.​

While this is a California law as of January 1, new labor laws are a national issue. There is a big gap between what full-time employees get from employers and what contractors get, from job protections, benefits and more. This law reflects a broader popular demand for lawmakers to do something to address that gap and to regulate gig work in new ways. New Jersey is already in the process of passing similar legislation to the California law, and the state of New York is considering legislation targeting gig work.

The California law misses the mark with a sort of "Make-America-Great-Again" thinking—not of course in the Trump/Republican sense. Rather than address distinct features of the modern economy, AB5 tries to force the world to conform to the models of the past. In particular, AB5 reinforces the two job classifications that already exist, employees and independent contractors, and makes it much more difficult for companies to use the latter. This fails to consider the features of the gig economy that make it different (both positives and negatives) and is worse for it.

The law and the legislator who championed it also don't seem to care if workers want alternative employment options additionally or instead. In response to stories on Twitter of freelancers and others losing their positions, San Diego-based Assemblywoman Lorena Gonzalez showed no doubts or sympathy, writing:

These were never good jobs. No one has ever suggested that, even freelancers.

​

B/ Turning point for how we think about higher education?

​Gallup released a new poll showing a sharp drop in the share of Americans who see a college education as "very important" and an uptick in those seeing one as "not too important":

The drop is even sharper for young people:

Hopefully, this opens up more discussion and action to improve higher education and expand the landscape of alternatives.

Hi there,

This week's note is the second part of my look at ghost kitchens, answering what do they mean for restaurants and the rest of the food delivery market.

I'd love to hear your thoughts on the notes so far! Drop me a note or add your comments and join the discussion of this week's post.​

If you know someone else would would like this, please pass it on.

​

A/ Ghost kitchens (II): Increased competition for restaurants and delivery apps

In the last post, I concluded that ghost kitchens will be good for customers and for some restaurant entrepreneurs.

But what do they mean for everyone else?

A helpful starting point is a simple model of the food delivery market today (pre-ghost kitchens):

Key points:

• Some restaurants still take delivery orders directly from customers and have their own delivery drivers, but the majority of the delivery market goes through the big app-based players (GrubHub orders are partly fulfilled by GrubHub drivers and partly by the restaurants)
• Competitive intensity is high in most parts of the value chain (although delivery is less competitive in some geographical markets)

Ghost kitchens will have three main effects:

• Reduce the overall costs of food delivery (as discussed last week)
• Make delivery more affordable for customers and increase the size of the delivery market
• Increase competitive intensity for restaurants and delivery apps

More affordable delivery, larger delivery market

Given the competitive intensity at each stage of the value chain, a large portion of the overall cost savings will translate to lower prices for customers. Lower prices will support a larger market. (Although to be fair, current prices may already be low due to competitive dynamics—in both cases I mean versus what otherwise would be the long-run prices and market size).

Increased competition for restaurants

Ghost kitchens will increase the competitive intensity for restaurants in several ways. Most notably, they will reduce the barriers to entry for starting a delivery-only restaurant and make it easier for successful formulas to expand. But they will also leverage their superior data (vis-a-vis their restaurant customers) to launch their own restaurant formulas, directly competing with restaurants.

From Venturebeat:

Underpinning most of these initiatives are large swathes of data that not only allow "food entrepreneurs" to know where to set up new kitchens but also tell them what to serve. Local demographic data can be used to figure out what kind of food outlets would work in a specific locale, while real-time consumer and operational data can be meshed to adjust menu options and test out new things.

For example (from the WSJ):

CloudKitchens also operates its own delivery-only restaurants in the commissaries, with names like Excuse My French Toast, Egg the F* Out, and B*tch Don't Grill My Cheese.

This is not good news for existing restaurants outside the successful ones previously discussed who can take advantage of easily adding more delivery capacity.

Increased competition for delivery companies

Although a larger market and lower costs are potential positives for delivery companies, they come at the expense of more competition. Ghost kitchens will make the delivery market more competitive in several ways:

• Ghost kitchens will make it easier for restaurants to operate across multiple delivery platforms, making it easier for customers to switch between apps
• By expanding the size of the market, individual geographical markets will be more able to support more competitors (i.e., for the same market share percentage, a larger market means a better cost structure through the ability to bundle deliveries into single driver trips).
• Concentrating the pickup location for multiple restaurants (and therefore orders) will also have a similar effect: higher delivery route density even at the same market share level. Depending on existing density in a given geography, this could benefit the market share followers more than the leader

Taken together, these work to reduce any winner-take-all effects in the app delivery market and increase competition.

What does this mean for ghost kitchen companies themselves?

The food delivery market is a large and rapidly growing one—and every other part of its value chain is highly competitive—so the ghost kitchen companies could be in an attractive position if they can carve out a sustainable competitive advantage. But that seems unlikely.

The benefits of the business model are captured substantially by the economics of a single facility, so the benefits of additional scale are somewhat limited. But there are some: Operating many facilities will provide some learning benefits, allowing them to build and operate the facilities more efficiently. The larger they are, the more valuable the data they collect will be. They will also be better able to serve larger chain customers.

The market for the services of ghost kitchens will also be relatively competitive. In cities (where the value of these facilities is greatest), the markets will be large enough to support multiple competitors. Less densely populated markets may be too small to support a second entrant but also are less attractive for the model to begin with. In general, the barriers to entry are mostly low (and will be more so as the model is understood).

So the competitive advantages of the ghost kitchen companies are likely to be weak. Still, they may end up with the least competitive part of a very competitive value chain.

Moreover, ghost kitchen companies can still create significant value for themselves in the role of developer, transforming low-value properties into high productivity uses (and McDonald's has shown a real-estate centered strategy can deliver financial success).

​Join the discussion with your comments​

​

B/ Sound bytes

​

1/ Ben Thompson reprises his view on the intertwined declines of TV advertising and mass brands

Global TV advertising sales have reached a major inflection point, contracting nearly 4% in 2019 despite global GDP growth of 3%. This is the "steepest drop since the depths of the economic recession in 2009" when the global economy contracted nearly 2%.

In his Stratechery update, Ben Thompson uses this news to reprise his astute view on the relationship between TV advertising and mass brands:

the very institution of television advertising is intertwined with the kinds of advertisers that use it the most, the products they sell, and the way they are bought-and-sold. And what should be terrifying to television executives is that all of those pieces that make television advertising the gold mine that it has been are under the exact same threat that TV watching itself is: the threat of the Internet…

The inescapable reality is that TV advertisers are 20th century companies: built for mass markets, not niches, for brick-and-mortar retailers, not e-commerce. These companies were built on TV, and TV was built on their advertisements, and while they are propping each other up for now, the decline of one will hasten the decline of the other.

​Read more on Stratechery​
​​

2/ Sam Lessin's take on a framework for the Future of Work

​In a post in The Information a few weeks ago, Sam Lessin laid out his working framework for thinking about the Future of Work:

Read more on The Information

But what do they mean for everyone else?

A helpful starting point is a simple model of the food delivery market today (pre-ghost kitchens):

Key points:

• Some restaurants still take delivery orders directly from customers and have their own delivery drivers, but the majority of the delivery market goes through the big app-based players (GrubHub orders are partly fulfilled by GrubHub drivers and partly by the restaurants)
• Competitive intensity is high in most parts of the value chain (although delivery is less competitive in some geographical markets)

Ghost kitchens will have three main effects:

• Reduce the overall costs of food delivery (as discussed last week)
• Make delivery more affordable for customers and increase the size of the delivery market
• Increase competitive intensity for restaurants and delivery apps

More affordable delivery, larger delivery market

Given the competitive intensity at each stage of the value chain, a large portion of the overall cost savings will translate to lower prices for customers. Lower prices will support a larger market. (Although to be fair, current prices may already be low due to competitive dynamics—in both cases I mean versus what otherwise would be the long-run prices and market size).

Increased competition for restaurants

Ghost kitchens will increase the competitive intensity for restaurants in several ways. Most notably, they will reduce the barriers to entry for starting a delivery-only restaurant and make it easier for successful formulas to expand. But they will also leverage their superior data (vis-a-vis their restaurant customers) to launch their own restaurant formulas, directly competing with restaurants.

From Venturebeat:

Underpinning most of these initiatives are large swathes of data that not only allow "food entrepreneurs" to know where to set up new kitchens but also tell them what to serve. Local demographic data can be used to figure out what kind of food outlets would work in a specific locale, while real-time consumer and operational data can be meshed to adjust menu options and test out new things.

For example (from the WSJ):

CloudKitchens also operates its own delivery-only restaurants in the commissaries, with names like Excuse My French Toast, Egg the F* Out, and B*tch Don't Grill My Cheese.

This is not good news for existing restaurants outside the successful ones previously discussed who can take advantage of easily adding more delivery capacity.

Increased competition for delivery companies

Although a larger market and lower costs are potential positives for delivery companies, they come at the expense of more competition. Ghost kitchens will make the delivery market more competitive in several ways:

• Ghost kitchens will make it easier for restaurants to operate across multiple delivery platforms, making it easier for customers to switch between apps
• By expanding the size of the market, individual geographical markets will be more able to support more competitors (i.e., for the same market share percentage, a larger market means a better cost structure through the ability to bundle deliveries into single driver trips).
• Concentrating the pickup location for multiple restaurants (and therefore orders) will also have a similar effect: higher delivery route density even at the same market share level. Depending on existing density in a given geography, this could benefit the market share followers more than the leader

Taken together, these work to reduce any winner-take-all effects in the app delivery market and increase competition.

What does this mean for ghost kitchen companies themselves?

The food delivery market is a large and rapidly growing one—and every other part of its value chain is highly competitive—so the ghost kitchen companies could be in an attractive position if they can carve out a sustainable competitive advantage. But that seems unlikely.

The benefits of the business model are captured substantially by the economics of a single facility, so the benefits of additional scale are somewhat limited. But there are some: Operating many facilities will provide some learning benefits, allowing them to build and operate the facilities more efficiently. The larger they are, the more valuable the data they collect will be. They will also be better able to serve larger chain customers.

The market for the services of ghost kitchens will also be relatively competitive. In cities (where the value of these facilities is greatest), the markets will be large enough to support multiple competitors. Less densely populated markets may be too small to support a second entrant but also are less attractive for the model to begin with. In general, the barriers to entry are mostly low (and will be more so as the model is understood).

So the competitive advantages of the ghost kitchen companies are likely to be weak. Still, they may end up with the least competitive part of a very competitive value chain.

Moreover, ghost kitchen companies can still create significant value for themselves in the role of developer, transforming low-value properties into high productivity uses (and McDonald's has shown a real-estate centered strategy can deliver financial success).

​

Hi there,

I hope you had an enjoyable Thanksgiving break!

Thanks for being an early subscriber! Please take a second to think if you know one or two other people who might like this too and forward it to them.​

This week's note covers:

• A look at the emergence of ghost kitchens (part one)
• The rise of niche brands and products in grocery stories
• Income share agreements, a relatively new model for educational financing that is gaining momentum
• Resources for building a successful marketplace business

​

A/ The rise of ghost kitchens: Plug-and-play facilities at scale for new delivery-only restaurants

One of the most exciting trends shaping the future of work is the democratization of scale across more sectors: new platforms are emerging that give small businesses almost instant-on access to infrastructure that takes advantage of much larger scale. This is creating new opportunities for small businesses and individuals.

Amazon Web Services (AWS) has done this for storage and compute infrastructure, enabling companies of all sizes to focus on their product or service without needing to build their own infrastructure. Shopify has done this for online stores, making it possible for new consumer businesses to focus on building differentiated products and brands.

Ghost kitchen companies aim to enable this same kind of plug-and-play access to scale for restaurants serving the app-based delivery market. As with AWS and Shopify, this has the potential to reduce costs for businesses and consumers, make it easier to start and operate a business in the space, fuel the expansion of downstream sectors and provide consumers with access to more variety and innovation.

Why is this important? The overall market for food is a big one: The USDA estimates spending on food (at home and away from home) represents about 10% of Americans disposable income, and app-based food delivery is rapidly growing share of this market. Restaurants are also a traditionally important category of small businesses. Data from the Small Business Administration indicates there are around one million of them in the US. Food delivery apps are already having a significant impact on the sector, and the arrival of ghost kitchens will have further effects.

In this week's blog post, I explain what a ghost kitchen is, highlight some of the players in the space, discuss the key reasons they create value and examine the implications for small business restaurant owners and entrepreneurs.

Summary points:

• Ghost kitchen companies are rapidly expanding in the US and internationally. Delivery companies are also getting into the business
• Ghost kitchens take advantage of several arbitrage and efficiency opportunities that reduce the cost of opening and operating a delivery-only restaurant. These benefits require a certain level of scale (not available to an individual restaurant) to capture fully
• For small business entrepreneurs considering starting a restaurant, they may offer a more attractive alternative to a traditional restaurant by reducing the cost and likelihood of failure
• For existing restaurants looking to expand, they offer a way to do this much more quickly and with less investment and financial risk than adding a traditional restaurant location

​
​Click to read the rest of the post on Remixing Work...​

​​

B/ "This is not your ordinary supermarket apple": the rise of new niche food brands and products

A combination of today's consumer preferences and changes in supply economics is fueling increasing variety and differentiation of brands and products to address more specific and narrow consumer segments.​

Several interesting pieces in the last week looked at how this is playing out in food and grocery. (Amusingly, two of them discuss separate innovations in apple varieties.)

1/ Planet Money's The Indicator discusses "Niche Products in Our Grocery Stores" in a short 10-min podcast, looking at why brands offer an increasing number of product varieties and how they do it (e.g., they say Kettle Brand Potato Chips 31 varieties!)

​Apple link​

​Episode site​

​

2/ A broader study on the subject, "The Rise of Niche Consumption," was published in September by Brent Neiman and Joseph Vavra, both of the University of Chicago.​

They find that, although our preference for particular things has increased as individual consumers, those preferences have diverged across consumers:​

"over the last 15 years, the typical household has increasingly concentrated its spending on a few preferred products. However, this is not driven by “superstar” products capturing larger market shares. Instead, households increasingly focus spending on different products from each other. As a result, aggregate spending concentration has in fact decreased over this same period."

​

3/ The first of the apples:The Indicator discusses how Washington State decided to develop and launch of a new variety of apple, the Cosmic Crisp.​

​Apple link​

​Episode site​

​

4/ The second of the apples--taking differentiation even further: In a longer discussion on new uses of internet video, a16z's Connie Chan and Avery Segal highlight how Chinese farmers are using short videos to market their produce directly to customers:

By selling directly to consumers instead of through grocers, farms can reach far more customers and earn more money. Meanwhile, viewers can support independent growers—meeting the face behind their fruit—and receive fresher produce than what’s available in local markets.​Ms. Feng, for instance, is a medium-sized fruit influencer in Sichuan province who broadcasts to 218K followers. Her orchard grows sugarcane and pomegranates, but she’s best known for “sugar heart” apples, named after the fruit’s heart-like appearance. This is not your ordinary supermarket apple.

C/ New education financing models are gaining momentum

Income share agreements (ISAs) are a different way of financing education. With an ISA, a student doesn't have to pay anything upfront or take out any loans. Instead, they have to repay a percentage of their salary for a certain number of years after completing their program. But they only have to pay if they have a job paying a certain salary level. If they don't find a suitable job or can only find one below the threshold salary, they don't have to pay anything.​

ISAs are being used by companies like Lambda School and Flockjay to fuel their popular bootcamp programs that help students learn skills and get jobs in computer programming and tech sales. They are also starting to be used by traditional higher ed institutions.​

Purdue University President Mitch Daniels has overseen the adoption of ISAs by hundreds of students at Purdue and published a column in the Washington Post last week, extolling their virtues and arguing for further adoption.​

The primary benefit to students is much better risk-sharing:

"An ISA is dramatically more student-friendly than a loan. All the risk shifts from the student to the investing entity; if a career starts slowly, or not at all, the student’s obligation drops or goes to zero. Think of an ISA as equity instead of debt, or as working one’s way through college — after college."

But an equally valuable systemic benefit is increased accountability for the educational institution and alignment of incentives with their students: if students are not graduating with the skills that make them attractive to employers and provide them with good-paying jobs, the school bears the cost.

This will lead to a positive loop for improving the ISA-financed programs. It will also lead to an increased level of transparency of outcomes than exists today. As more and more people choose these programs to advance their learning and career switching or advancement objectives, it will also force many traditional programs to adapt to become more efficient and effective if they want to be able to continue to attract students.​

As an interesting side note, ISAs are a relatively new innovation as a product, but the idea for them is not. Milton Friedman suggested the appropriateness of this kind of equity investment to support education in a 1955 paper.​

The takeaway: ISAs are a welcome innovation in a sector that needs it. They share risk better, align incentives and drive accountability. They offer students what will sometimes be an attractive alternative to traditional education financing and are also enabling new, promising education and training models.

​

D/ Resources on building successful marketplace businesses

New marketplaces are a huge factor in shaping the future of work. Before the rise of Uber and others, the concept of a "gig worker" didn't exist. Now three in 10 adults in the US work in the gig economy according to the Federal Reserve.

Below is some great content from recent weeks that will be useful to anyone working at a marketplace business or thinking of starting one and anyone more generally interested in how they operate.​

1/ "How to Kickstart and Scale a Marketplace Business"

Lenny Rachitsky has a fantastic four-part series in his Lenny's Newsletter on "How to Kickstart and Scale a Marketplace Business: 🐣 Crack the Chicken-and-Egg Problem."​

It compiles the learnings from the people who built "today's biggest marketplaces, including Airbnb, DoorDash, Thumbtack, Etsy, Uber and many more."

• ​Part 1: Constrain the marketplace 🔬​
• ​Part 2: Decide which side of the marketplace to concentrate on 🧐​
• ​Part 3: Drive initial supply 🐥​
• ​Part 4: Drive initial demand 👋​

2/ Podcast episode: VC perspectives on marketplaces

Sarah Tavel, partner at Benchmark, and Nabeel Hyatt, partner at Spark Capital, discuss their experience investing in marketplace businesses on Erik Torenberg's Venture Stories podcast

​They cover a lot of ground.

One takeaway is the importance of liquidity over growth in a marketplace's early days. (Although, of course, liquidity requires some level of growth.)

But as an example what this means if you're Uber just starting out, it's far more important to focus on building a core market in which riders can get picked up a very high percentage of the time within a few minutes than it is that you achieve a certain level of growth in number of total rides or revenue.

​Apple link​

​Episode link​