November employment growth slows in both tangible and intangible industries

Employment growth slowed in November according to the BLS data released this morning. Nonfarm payrolls were up by only 210,000 employees, compared to 546,000 in October and 379,000 in September.

Employment rose in both intangible-producing and tangible-producing industries at a slower pace than in previous months. Employment in intangible-producing industries grew by just 91,300 while employment in tangible-producing industries was up by only 118,200. This compares to increases of around half a million in both tangible-producing and intangible-producing industries during the past summer.

Interestingly employment in the tangible portions of the education and health care sectors (Nursing & Residential Care Facilities and Child Day Care Services) actually declined while employment in the intangible portions increased (but at a slower rate). Employment in Personal & Laundry Services, Telecommunications, and Government also declined. One of the few industries to see an increase in employment compared to last month was Tangible business services, due to higher than last month’s employment in Services to Building & Dwellings and the Postal Service.

As I have noted in earlier postings, the labor market seems to have settled back into the post-Great Recession, pre-pandemic pattern of relatively equal growth in tangible-producing versus intangible-producing industries – but at a slower rate. The COVID-19 pandemic has done little to disrupt to dramatic shift in the tangible-intangible structural balance that emerged after the Great Recession.

For more on the categories, see my explanation of the methodology in an earlier posting

Overall knowledge-related business investment continued to grow in 3Q 2021

But the semiconductor shortage may be creating problems for investment in information processing equipment

As BEA’s data released yesterday showed, GDP growth in the 3rd quarter of 2021 came in at a disappointing 2% annual rate. This was well below the 6.5% growth rate of the previous quarter and below general expectations. However, business (non-residential fixed) investment in knowledge-related areas grew by 5.7%. This grow was due to healthy increases in investment software (up by almost 15%) and R&D spending (up by over 9%). Investment in information processing equipment unfortunately declined by 6%. Total business investment in all other areas declined by 3.5%.

The decline in investment in information processing equipment is especially worrisome. This marks the second quarter in a row of declines following four earlier quarters of growth. This may be a reflection of the ongoing semiconductor shortage.

Another casualty of the computer chip shortage is the auto industry. Both expenditure on motor vehicles and parts and investment in transportation equipment dropped dramatically in the 3rd quarter. That decline accounts for much of the overall slow growth in GDP. If motor vehicle expenditures had simply stayed at the 2Q level, the overall GDP growth rate would have been double, at a respectable 4%.  

Knowledge related business investments did not suffer as great a cutback as other business investments in the COVID-19 slowdown and have been growing since 2Q20. They now account for 58% of total business investment (up from 50% in 3Q19). Looking at only the two digital-related investments of information processing equipment and software, this subcategory makes up 40% of business investments.

[Note: I define knowledge-related investment as the combination of investment in Information Processing Equipment, R&D, and Software. The first of these three categories is reported in the GDP data as a subcategory of Non-residential Fixed Investment: Equipment. The latter two are reported as subcategories of Non-residential Fixed Investment: Intellectual Property Products.]

Lessons from agriculture on going beyond technological innovation

One does not normally think of technological innovation and agriculture. But I recently came across an article in PwC’s Strategy & Business on “The fourth industrial revolution in agriculture” from a couple of years ago. The authors, Sebastiaan Nijhuis and Iris Herrmann, describe how technologically sophisticated agriculture has become and how agribusiness is going about implementing new technologies. These technologies range from AI to track and better manage cows for more efficient milk production to the use of drones and IoT sensors to improve crop yields.

However, what really struck me about the article was not the new technologies. Rather it was their argument on the need for organizational and strategic change to better utilize the technologies. And, in turn on how the development of these technologies will force those changes in ways we might not expect. For example:

“One firm is developing a swarm of miniature autonomous robots that can plant seeds. Controlled by a farmer’s handheld tablet, which is operated with the help of satellites and cloud-based software, the swarm will be able to put each seed in the right place with greater precision than current approaches can. Not incidentally, the technology will eliminate the need for planter bars, tractors, and tractor operators.”

They go on to note that:

“The most common response of companies has been to plug new technology into old business models, with the hope of enhancing those models with smarter tools and more data. But that tactic is flawed. Making old models work better isn’t enough — not when technologies are enabling all-new models that can render the old ones obsolete.

Many pesticide and fertilizer companies, for example, are using 4IR [4th Industrial Revolution] technologies to provide better products and roll them out faster than before. That might sound like a success story, but precision farming — which uses IoT sensors, high-resolution 3D aerial imagery from drones, and AI-powered analytics to analyze the characteristics of soil and the behavior of crops down to the square inch — may soon significantly reduce the need for fertilizers and pesticides altogether.

A better approach for those manufacturing companies is to discover and develop these new business models, creating new markets along the way. Instead of looking for a better product, companies should look for better solutions for the problems that their customers face, whether those customers are farmers, agricultural suppliers, or end consumers. Many successful solutions will bring together products and services from multiple companies, rather than just using products manufactured by the solution provider.”

Good advice in general – not just for agribusiness.

More on disclosure of human capital data

One of the most important policy steps the US could take to heighten awareness of intangible assets is to require companies to include them in companies’ financial statements (see earlier postings). In a speech given earlier this summer, SEC Chairman Gary Gensler reiterated his support for enhanced disclosure by companies of their human capital:

Further, investors have said that they want to better understand one of the most critical assets of a company: its people. To that end, I’ve asked staff to propose recommendations for the Commission’s consideration on human capital disclosure.

This builds on past agency work and could include a number of metrics, such as workforce turnover, skills and development training, compensation, benefits, workforce demographics including diversity, and health and safety.

Disclosure helps companies raise money. It helps the efficient allocation of capital across the market. And it helps investors place their money in the companies that fit their investing needs.

As the SEC works through its process, it is important to understand what companies are already doing. A new study out from the analytics organization JUST Capital (The Current State of Human Capital Disclosure in Corporate America: Assessing What Data Large U.S. Employers Share) provides some helpful insights.

The report identifies 35 human capital metrics (28 of which they were able to collect data). The metrics cover six themes: employment and labor type; job stability; wages, compensation, and benefits; workforce diversity, equity, and inclusion; occupational health and safety; and, training and education. The specific metrics are very detailed. For example, the employment and labor type theme includes not only the number of employees but the number of on-site contractors and on-site temporary or seasonal workers. The job stability theme includes the amount of voluntary turnover by gender and race/ethnicity. The wages theme includes the minimum wage to local minimum wage ratio.

Their analysis of the data collect from 100 of the largest U.S. companies shows why SEC action is needed. To start with, disclosure of human capital data is low. The disclosure rate is below 20% for the majority of metrics. The most commonly disclosed metric was the number of full time employees; even there the disclosure rate was below 40%.

And even then when the data is disclosed, it is likely to be included in a Corporate Social Responsibility Report, Sustainability Report or other Impact Report rather than in the company’s financial report (10-K). As important as these reports are, the study points out that they are optional publications with no set standards for reporting and no auditing requirements. Thus, they are difficult to use to make comparisons across companies and may lack accountability.

The report explains why this lack of data standardization matters: “Without the ability to compare how companies are performing on various human capital metrics, investors can’t make well-informed decisions about where to direct their holdings, potential hires can’t factor a company’s treatment of workers into their employment choices, and customers can’t shift their purchasing practices to support companies leading the way.”

Likewise, the same need for comparability is the reason for why such data disclosure must be mandatory. I understand that some object to increased mandatory reporting, arguing that it gives away important information to competitors. Remember that there was a time when the disclosure of even basic financial data such as revenues, expenses and profits was opposed on the grounds it was proprietary. But as I noted back in 2005 in our paper on Reporting Intangibles, without such information investors and others are flying blind.

Managers also need better data. The JUST Capital study l0oked at public disclosure of human capital metrics. I wonder how many companies don’t even collect such data. Mandatory disclosure may force companies to do what they should be doing anyway: collecting data to better understand and manage their intangible assets including human capital.

So, the sooner the SEC moves to requiring more disclosure of data on human capital, the better. And thanks to JUST Capital for continuing to push on this topic.

Importance of local assets

Over the years, one of the leading themes of the Intangible Economy has been the need for communities to build upon the local assets. So, I am always on the lookout for interesting examples.

One such example comes from Sparta, Greece. Petros Doukas, the Mayor of Sparta, describes what they are doing to build on local assets. They are deliberately not trying to become a “high tech” center. Instead, they are looking at areas where they have some type of existing advantage and have identified the sectors of agriculture, tourism, cultural activities, and sports. You can see his remarks describing what they are doing in each of these areas at this YouTube video starting at about minute 16:15

[FYI – This clip is part of an excellent discussion on local development as part of the Framing the Future series organized by the Global Federation of Competitiveness Councils (GFCC) (click here to see the links to previous sessions).]

Sparta is an example of the economic gardening approach to local development. And as I pointed out in an earlier posting (“A New Narrative for Rural America”, Knowledge Management as an Economic Development Strategy and “Building on Local Information Assets”), communities following this strategy adopt a “build, not buy” approach to growing local businesses. 

That is not to say that efforts to create technology and innovation clusters are not important. They are. But as Mayor Doukas points out, many communities would be better off adopting a lower tech strategy. In any event, deciding which way to go starts with understanding your local assets.

Thinking about manufacturing and services

One of the ongoing themes of my work has been the restructuring of the economy including the fusion of manufacturing and services. I have long argued that these categories are not useful in helping to understand economic activity. There are some cases where the categories are clear. For example, haircuts are clearly a service. But increasingly the boundaries are fuzzy. So-called manufacturing companies are more and more selling their products as a service and service activities are an increasing part of value-added.

And sometimes, services want to be seen as manufacturers. A story in today’s New York Times on the public-private revolving door for tax lawyers and accountants reminded me of how far and how creatively services can be considered “manufacturing.” Before it was repealed in 2017, Section 199 of the tax code provided a tax deduction for “domestic production” aka manufacturing. While revenues from retail food preparation were explicitly not allowed to take the deduction, creative interpretations of “production” were accepted including the notorious cheesecake-slicing-as-production claim mentioned in the Times story. And there was a florist who claimed the manufacturing deduction as they were using individual flowers to produce bouquets.

Interestingly, the law specifically included engineering and architectural services as qualifying activities. I assume these services were seen as important adjuncts to construction (itself a qualifying production activity).

I should note that the opposite classification of construction as services or manufacturing is used by the Institute of Supply Managers, which includes construction in its services index. I suspect that this is a legacy of the original formulation of this as the index of non-manufacturing, as opposed to their index of manufacturing (see earlier posting).

These stories about manufacturing versus services should give us pause. Admittedly, the cheesecake example is somewhat silly (of course not to those who got the tax break). But if assembling a car from various parts is considered manufacturing, isn’t assembling a bouquet from various parts also? If testing a computer chip is considered part of the manufacturing process, what about chip design services? And then there is the thorny question of these activities being carried out by outside firms (considered a service) rather than in-house (considered part of manufacturing).

We need a serious effort to rethink how we envision the economy. Part of that is embracing the idea of making value as opposed to making things (see earlier posting). But it also includes restructuring our industrial categories – possibly around types of end output (e.g., food, shelter, transportation, health, entertainment, etc.). All of the recent discussions about supply chains indicates how interconnected economic activities are and how these activities cluster into groups.

I’m not smart enough to come up with the best framework. I hope someone is. Otherwise, we will continue to make economic policy based on a view that is clearly out of date.

Innovation Lessons Learned From Failure

Studies of innovation usually focus on success stories – what worked and therefore what others should do. The National Science Foundation (NSF), specifically the National Center for Science and Engineering Statistics (NCSES), has recently released a study that takes the opposite approach: what can we learn from failures. Carried out by team of researchers from Center for Innovation Strategy and Policy at SRI International, the study (Understanding Unsuccessful Innovation) looks at 19 case studies of failures (see list below) and pinpointed the root cause of the failure.

Of course, it is somewhat problematic to draw conclusions from such limited sample of case studies. But they do provide an illustration of the various ways that an innovation can fail. And it should be noted that not all of these “failures” ultimately turned out to really fail. Some of these pivoted to a different market use that originally conceived (often much more specialized and narrower than the original target market).

Thus, the framework developed is probably more important a finding than the statistical evidence provided.

The framework identifies five root causes of innovation failure:

No market demand. A product or business process was developed and brought to market, but there was little market demand for it.

Poor performance. A product or business process was developed and brought to market, but it failed to function as intended.

Insufficient complementary assets. A product or business process was developed and brought to market, but the adjacent business inputs required for its successful customer use were not sufficiently available.

Poorly defendable position. A product or business process was developed and brought to market and met with positive market interest, but the innovator was unable to secure the innovation’s market position, for example, by way of intellectual property protection.

Regulatory restriction. A product or business process was developed and brought to market, but regulatory restrictions on its use limited its economic value.

Note that the study includes a breakdown of the “no market demand” category into innovations that are not useful, price point that are too high for the market, failure to meet investment expectation, and wrong market targets.

Those failures can manifest itself during one of the three time periods of innovation: launch, growth, or maturity. The root cause failure, however, occurs one of the earlier 5 stages: ideation, product development, launch, growth, and maturity. The model thus is as follows:

The model helps pinpoint where the fail occurs, rather than when it occurs.

For example, the paper concludes that iTunes Ping failed during launch due to poor performance that was the result of failure in the product development stage (specifically not having a data-sharing agreement with Facebook in place). Hoverboards failed in the transition to maturity due to a poorly defendable position in the growth stage. Google Wallet failed in the growth stage due to insufficient complementary assets that should have been addressed during product development.

The Google Wallet case study (along with the Wii U, Google Glass, Segway, Iridium Satellite Phone, and other cases) points out one of the limitations of the study. Each of these innovations are still in use – either as part of another innovation (Google Wallet as part of Google Pay; Wii U as part of Nintendo Switch) or as a scaled back version (Google Glass, Segway, and Iridium Satellite Phone). Over half of the cases resulted in this pivoting of the innovation to new markets and/or new uses. A more complete study would look in greater depth at the process of pivoting from apparent failure to at last limited success.

Another concern with the study is its focus on a sole root cause of failure. It seems to me that the root causes are not mutually exclusive. For example, the failure of Segway seems to be a combination of no market demand (due to a high price point) and regulatory restrictions. A look at the multiple causes, and their interactions, would strengthen the model. Especially important is how the lack of complementary assets can create low market demand and poor performance.

I come away from reading the case studies with three insights. [Note that in keeping with my concern over the small sample size, my comments are based more on the model and the insights from the case studies rather than statistical analysis of the case studies.]

The first insight concerns the failure to meet a market demand. It should come as no surprise that no market demand, especially not useful, is a root cause of an innovation failure. Yet, most of our public policy seems to ignore that point. All innovations are seen as beneficial. Innovations fail because of the lack of funding or some other barrier to adoption – not because they are not useful. I call this the “electronic swizzle stick” syndrome:  just because we can build an internet-connected battery powered swizzle stick that can stir your cocktail in the perfect manner does not mean that there is a market need for such an item (although I keep watch to see if such a product appears in the airline in-flight catalogs). The Juicero case illustrates this point.

The second insight is that all of the causes of failure can be addressed somewhere in the innovation process. But the causes of failure need to be identified early and steps taken to avoid the failure. As the saying goes, an ounce of prevention is worth a pound of cure.

This maybe the hardest insight to operationalize. Such proactive measures do not come easily. The whole fail-fast ethos is built around the notion of rapid learning rather than pre-emptive action.

Fortunately for the rapid learning approach, the third insight from the case studies is that failure do not mean failure. As noted earlier, many of the innovations were at least partially implemented – just maybe not in the way or to the extent originally envisioned.

The study does not delve into policy proposals. As befitting its sponsor NCSES, the study looks at the implications of the model for statistics on innovation collected as part of the American Business Survey, and at issues for future research. Let me therefore speculate on some policy implications.

I believe there are public policy programs already in place to address the innovation failure. Directly targeted to the issue of market demand is the NSF’s I-Corps program (and its many offshoots). The I-Corps process forces innovators to confront real world applications and market needs. The Manufacturing Extension Partnership (MEP) and the Small Business Innovation Research (SBIR) – Small Business Technology Transfer (STTR) programs exist to address technical issues of poor function. But I-Corps reaches only a limited audience and MEP and SBIR/STTR programs are narrow in focus.

Let me suggest a more “rough and ready” approach. As part of any assistance to companies’ innovation activities a checklist could be consulted. It would a rudimentary checklist just to get people thinking about the possible problems just as the I-Corp process forces innovators to think realistically about the market needs. I realize that some may claim that this is just setting up another series of barriers to innovative activities – the creation of a way to say “no.” But it could be useful overcoming barriers by addressing the failure points proactively. And it could also force the naysayers to explain their objectives.

Obviously, such a checklist would not be a silver bullet to ensure that all innovations succeed. But it might be a small step toward helping prevent more innovations from failing due to a lack of foresight. Worth a try at least.

Case studies

  • Google Glass
  • Microsoft Windows Vista
  • Samsung Galaxy Note 7
  • iTunes Ping
  • DVD-Audio
  • Hoverboards
  • Boeing 737 MAX
  • Zozosuit
  • Juicero
  • Google+
  • Ubuntu Phone
  • Sedasys
  • Google Wallet
  • Segway
  • Iridium Satellite Phone
  • 3-D Television
  • Wii U
  • Laundroid
  • Sony BMG Extended Copy Protection (XCP)

August employment grows – but slower and differently for intangibles and tangibles

Breaking the pattern of the past few months, August’s employment data slows the clear effects of the COVID-19 Delta virus with a much smaller than expected increase of 235,000. As I have noted in an earlier posting, the labor market had been settling back into the pre-pandemic pattern of relatively equal growth in tangible-producing versus intangible-producing industries. August changed that.

In August, employment in intangible-producing industries rose much faster than in tangible-producing industries – admittedly at slower rate of 189,000 in intangible-producing industries versus 46,000 in tangible-producing industries. This compares to increases of around half a million in both tangible-producing and intangible-producing industries in the past few months.

The most dramatic shift was in Accommodation and Food Services which went from strong growth over the past few months to actually declining in August. But the slowdown in employment growth was generally across the board (see chart below). Only Professional and Business Services saw a real increase in employment growth in August.

I doubt the August data signals a return to the pre-2010 situation of rapid growth in employment in intangible-producing industries versus in tangible-producing (see earlier posting). But it does constitute a break in the parallel trajectories of the intangible-producing and the tangible-producing industries.

For more on the categories, see my explanation of the methodology in an earlier posting

Productivity, Organizational Structure and Work-From-Home

Recently, there has been a great deal of discussion about the effect of the pandemic on productivity. Has the shift in work, especially the increase in work from home (WFH), boosted productivity? Will a return to the office lead to more or less productivity? Or, to put it another way, will the return to the office enhance or degrade the key intangible asset of organizational competence?

Articles by David Rotman at MIT’s Technology Review, Heather Long in the Washington Post, and Neil Irwin in the New York Times paint a positive picture over the overall trend in productivity.

But, as Chip Cutter points out in the WSJ, companies are struggling with how to reopen the office that is not disruptive. 

The answer to the productivity question depends on the interaction between organizational changes and new technologies. Most of the productivity optimism is based on an expectation of continued deployment of advanced technologies, specifically artificial intelligence, machine learning, automation, and robotics. But experts such as Erik Brynjolfsson have long argued that concomitant organizational changes are need for deployment of these technologies (for example, see here and here).

For the most part, these changes in organizational structure are needed to take advantage of the productivity-increasing features of the new technology. The classic example is the change in factory layouts due to the shift from steam-driven belt and pulley system to individually electric engine driven machines.

But equally important is the role of a shift in work organization in facilitating the acceptance of new technologies. The pandemic-induced forced shift to WFH seems to have broken down barriers to the deployment of digital technologies. As a recent large survey of companies by Barrero, Bloom, and Davis concludes, “the pandemic created the conditions for coordinated experiments with WFH in networks comprised of firms, customers and suppliers, yielding lessons and know-how that were hard to acquire beforehand. In sum, the pandemic swept aside inertial forces related to experimentation costs, biased expectations, and coordination within networks that had previously inhibited remote work.”

The challenge facing companies now is how to harness these productivity drivers. Can they take advantage of the window of opportunity of lower barriers to technological change and use the WFH experience to create the new organizational structures needed to operationalize the new technologies? Or will forcing workers back to the office reimpose the pre-pandemic organizational structure and negate any organizational learning?

My guess is that we will see a mixed result. We are in the middle of a huge social experiment. Researchers, managers, and workers will need to monitor the experiment closely for lessons-learned and new “good practices.”

As a side note, I find the ongoing discussion on the impact of the elimination of the work commute on productivity to be somewhat confusing. It has been argued that the shift to work-from-home has raised productivity. For example, the Barrero et al. survey cited above concludes that WFH will increase productivity by 4.8%, half of which is due to reduced commute times.

But this is a case of working longer, not working smarter. Official productivity measures do not include commuting time as work time. Adding in time previously spent commuting increased the total number of hours worked. Working more hours presumably increases production (the amount of work you get done). This does not necessarily increase labor productivity (as measured by output per hour). And if the result of working from home doesn’t lead to a measurable increase in output (a distinct possibility for office work), then productivity actually goes down (more hours, same output). Also, any reduction in commuting time is a one-shot improvement (assuming that the number of people working from home does not continue to increase) and may be reversible (as workers are required to go back to the office at least part-time). And of course, it does increase company profits if workers are not paid more for those extra hours – and likely they are not (more output due to more hours for the same amount of expense).

July employment shows continuing growth

This morning the BLS announced that employment in July increased by 943,000 – more than expected. And the unemployment rate dropped to 5.4%. Employment growth was across the board. As in previous months there was a continued resumption of economic activity in two areas which have direct public contact. Employment in Accommodation & Food Services was up by 2.6% and Arts, Entertainment & Recreation employment was also up 2.6%. Only Tangible Education and Health Services saw a decline.

As I mentioned last month, the data continue to show the economy settling back into the pre-pandemic pattern of tangible-producing versus intangible-producing industries of the last decade.

From 2000 to around 2010, employment in tangible-producing industries slowly declined while employment in intangible-producing industries rose. And as a result, the share of total employment in intangible-producing industries passed that of tangible-producing industries some time in 2009. This was the continuation of a long trend in the growth of the intangible economy.

But around 2010 something happened. Employment in tangible-producing industries started growing at about the same rate as employment in intangible-producing industries. And the split between the two in terms of percentage of total employment stabilized. The pandemic reversed that trend with employment in tangible-producing industries dropping much faster than in intangible-producing industries. We now have enough post-crash data to clearly see that the 2010-2020 trend of equal employment growth is reasserting itself.

So, what happened in 2010? I suspect that the long-awaited Information Society (or Post-Industrial Society if you prefer the older title) finally arrived and with it a change in the tangible producing process. The nature of the output between tangible and intangible may be different, but all production processes are becoming intangible-heavy. For example, manufacturing is now a knowledge-based activity. Another explanation may be the fusion of tangible and intangible output (often referred to as “servitization”). Companies no longer sell just tangible products but combine the physical with an intangible service (such as home alarms).

The pandemic lockdown disrupted the economy. But it did not fundamentally alter this shift.

For more on the categories, see my explanation of the methodology in an earlier posting