John Haltiwanger
Subject: Surging Business Formation
Bio: Professor of Economics at the University of Maryland
Transcript:
Larry Bernstein:
Welcome to What Happens Next. My name is Larry Bernstein. What Happens Next is a podcast which covers economics and politics.
Today’s topic is Surging Business Formation.
Our speaker is John Haltiwanger who is a Professor of Economics at Maryland and is a specialist in new businesses. I want to hear from John about the growing importance of new firms to increase productivity and provide job growth. In addition, I want to learn about how the movement towards self-employed and 1099 independent contractors will influence the job market.
I recently held a conference in Washington DC with a bunch of my friends where we chatted with John. This podcast will be different from normal because some of the questions will be asked by my friends.
John, please begin with your opening six-minute remarks.
John Haltiwanger:
There has been a surge in business formation in the United States in the pandemic that is continued through today.
There's data at the Census Bureau that is released monthly. It is real-time applications for new employer identification numbers. And you might say, okay, well how important is this bureaucratic indicator of individuals applying for new EINs that are linked to new employer businesses? We care about new employer businesses because the evidence shows their important sources of job creation, innovation, productivity, and growth. If you see a surge in non-employers that is suggesting lots of people are going out on their own.
We have seen a huge surge in the pandemic, and it has even been larger for these new non employers. We are seeing new jobs, a million per quarter from these establishment births, and this is much higher than in 2019. A new firm birth, it is a brand-new greenfield entity.
Who are these guys? What are they doing? What is their contribution to the economy? And the answer is, we do not know yet, but I am happy to speculate with you. Historically it takes, a while to figure out whether a new business is going to be successful or not. If a new business starts at five employees and ends up at 50, that is an enormous amount of job creation. And if we are getting hundreds of thousands of those that is the contribution that I am talking about here.
Using just New York to illustrate this, and it is true in every major city with over a million in population is the surge in EIN applications. The surge in new businesses is not in Manhattan, it is in the surrounding areas. And we find that it is associated with a wide range of support businesses, things like food and accommodations, gyms, and the like. And what do we think is going on? What is happening with working from home, the daytime population is spending time in a different place than they were in 2019.
We have seen a surge in businesses in the surrounding areas to support that new working from home. It is a reshuffling of the geographic location of activity. It is in places like food and accommodations.
High tech startups had been on a decline since the Dot-Com boom from the 1990s and early 2000s. They were still out there, just that we were not seeing quite the surge that we saw in the 1990s. So, what is strongly associated with this enormous surge in high tech? AI shows up repeatedly in the write-in information. There has been a takeoff in AI startups from these EIN applications.
Larry Bernstein:
Our previous podcast was with Betsey Stevenson who is the former Chief Economist at the Department of Labor. She spoke about the two major employment surveys that the BLS produces each month. These surveys are getting radically different numbers for the current state of the employment market for the past 12 months. The Household Survey showed almost no employment growth while the Establishment Survey showed modest growth of 2.5 million jobs for the year.
Both the Household and the Establishment Surveys have their problems. How can we figure out the truth about the state of the US job market?
John Haltiwanger:
Lots of interesting measurement questions here. The numbers for employment from the tax data not from either survey is the truth. Both the BLS and the Census bureau get very high-quality tax data. The establishment births is from every state and gets their reports from the UI account system.
Larry Bernstein:
The UI account system is the state generated unemployment insurance numbers. And do you think the unemployment insurance data is correct?
John Haltiwanger:
Those are highly accurate. So, the BLS uses it to do the benchmark revision. That got so much attention recently because it was so large. I think the truth is the tax data.
Larry Bernstein:
The US employment data is the most important economic information that is released each month for understanding both the US as well as the world’s economy. It moves markets.
The two employment surveys that the BLS produces each month. The first is the Household Survey that uses a panel of households, and the government calls 60,000 of them each month.
The survey panel is called the Current Population Survey or CPS and they may not be picking up the phone. And we know from political polling that creates a selection bias.
In the Household Survey, the government asks if anyone living in the house got or lost a job in the past month. The second survey is the Establishment which gets job payroll data from America’s established firms. Which survey do you think is closer to the truth, and what are the problems with them?
John Haltiwanger:
I put a much higher weight on the payroll survey. Why? Unfortunately, even though the CPS is a longstanding hallmark of the US statistical system. Household Survey response rates on the CPS have been falling quite steadily over the last 10 or 15 years and accelerated in the pandemic.
It is not a crisis in the United States, but what's happened in the UK. The UK's CPS response rates fell to literally 20% that the Office of National Statistics in the UK stopped publishing the monthly unemployment rate. And that created chaos in markets. I do not want to be quoted or suggested that I think the CPS is in as much trouble as the ONS, but response rates are especially low on household surveys. They are also falling on establishment surveys.
The payroll survey, which is a much larger survey, 300,000 establishments, response rates are falling but not nearly at the same rate as the Household Survey.
Larry Bernstein:
The Establishment Survey is at its best when it calculates job gains and losses for America’s largest firms. But most employees work for smaller companies and the real action is happening with newer firms that do a lot of hiring.
The Establishment Survey uses a model to make an estimate for births and deaths of new companies and then guesses their hiring practices. It turns out that this is hard to do and prone to error especially when the labor market is turning.
John Haltiwanger:
The payroll survey does probabilistically sample all size classes of firms. So, it is doing that. Now having said that, you are still on exactly the right track because here's what the stat agencies are stuck with and that's why you need to be appropriately cautious about all the monthly high frequency series.
The establishment surveys, so what do they do, because response rates are variable. They take the establishments that reported last month, they take the establishments that reported this month, they take their growth rate and then they build the statistics based off that. So, it is going to do a terrible job with births and death.
When I said they do sample small establishments, it takes them a couple years indeed to get the new births that they are picking up in the tax data.
I agree with you completely, the birth death model that BLS uses is quite weak.
Larry Bernstein:
Each year, the BLS does a revision of the Establishment Survey employment statistics. In their most recent revision, they discovered that they had overestimated the number of new jobs for the previous year by 800,000 jobs or by 50%! What do you make of this error and the benchmark revision?
John Haltiwanger:
So, it might freak you out when there's these large benchmark revisions, but the good news is that there is a benchmark revision and there is not for the household survey.
Larry Bernstein:
Each year, there are millions of illegal immigrants enter the United States who are working age. Do these immigrant workers get properly included by the Household Survey? I cannot imagine that they pick up the phone and answer these questions from the government. I also would be surprised if the immigrants work for large established firms. And I suspect that they may not fill out their taxes to be included in the tax data?
John Haltiwanger:
If we were in Brazil or Columbia, this issue you are talking about is overwhelming with the informal economy, workers off the books, represents like 50% of the labor force employment. That has not been true historically in the United States. Now are we creeping in the direction that is a good question.
Larry Bernstein:
Earlier you mentioned that you think that the tax data is closest to the truth. What happens when you match up the individuals who are in the Household Survey with their respective tax filings?
John Haltiwanger:
We started to match up the tax data literally for people filing their unemployment insurance taxes, their payroll taxes, their taxes to the IRS to the Household Survey. And we have noticed enormous discrepancies.
The Household Surveys are especially bad at picking up individuals who are doing work but it's not their primary activity or they're only doing it on a part-time basis. It's especially bad for the self-employed, multiple job holders, or individuals using work in a stop gap way. Uber drivers and the like. How important is that? Well, there is this enormous fringe in the terms of their attachment to the labor force that are engaged in work activity.
And even if they are not working that much themselves in any given month it adds up to millions of such workers. It matters, okay? And the Household Survey is just terrible at picking that up. Low response rates, poor capturing of marginal work activity.
Larry Bernstein:
What about people who are self-employed?
John Haltiwanger:
We literally matched up all the individuals who report to the IRS that they are engaged in self-employment activity and it's a huge number. And we could find all of them in the CPS. And then we asked what fraction of the individuals who report to the IRS that they have self-employment activity that they report to the CPS that they have self-employment. It turns out 70% of self-employed do not report to the CPS that that's what they're doing. So, all due respect to household surveys, I think they are in trouble.
We need to find alternative ways of capturing the statistics that we are getting in the Household Survey. The payroll surveys also need work for the reasons you articulated, like the birth-death model is a big deal. They could do a better job. If I had to say which surveys are in bigger trouble, it is the Household Survey.
There is just an enormous amount of mismeasurement. Here's why, unemployment itself is a complicated concept. You had to both say you were not working and then you had to explicitly say I have not looked for work in the last four weeks.
People do not answer that question very well.
Larry Bernstein:
Just for clarity, the BLS defines someone who is unemployed as being someone who is not working but only if that individual has looked for work over the past four weeks. What else do the government employment statistics get wrong.
John Haltiwanger:
A lot of people who classify themselves in the CPS as students say they are not working. They are working on the side. People who say they are retired have employment. A lot of people who say that they are staying at home to take care of kids are making money. Okay, now is it their primary activity? No, the CPS is its best trying to figure out your primary activity. In terms of secondary activity. It is terrible.
Larry Bernstein:
It seems were asking too much from people in answering questions about their working lives. Our jobs do not fit nicely into the buckets that the Bureau of Labor Statistics wants to put us in. And furthermore, the nature of work is changing dramatically.
On a previous podcast, Dan Altman who is the former Chief Economist at Instawork explained that the company matches up jobs and workers who want short-term employment on their website. The types of short-term employment that is offered is typically in a warehouse, a restaurant, a convention, or a special event. What Dan Altman found surprising was that workers do not follow a 40-hour 9 to 5. Instead, some workers work an 80-hour week and then take the next couple of weeks off. Most workers take jobs in one industry like working in a warehouse one week and then work as a bus boy the next. Some workers will follow the job market geographically. They work at Formula One in Austin one week and then in Miami Beach at Art Basel a couple of weeks later. This work behavior does not fit in well with our government statistics, but it is fabulous in that workers are employed and offer choice and opportunity. These workers are 1099 contractors or self-employed who work in combination with other people.
John Haltiwanger:
I have been a little harsh on the surveys. The good news is the stat agencies are aware of these issues and are working on them. These are hard problems.
People figured out partly due to the pandemic, it was a triggering event that you could go out on your own and contract out in rich ways that's part of it.
On a monthly basis, hundreds of thousands of individuals fit into the category that you are talking about. They are bouncing across activities as employees. They are also bouncing across businesses as independent contractors. The labor market's surging even more than you think it has because of all the activity that is exceedingly difficult to pick up in both household and payroll surveys.
What do we do about it? The agencies are increasingly recognizing they need to start tapping into alternative data sets. Maybe track digitized work activity. The agencies are increasingly doing that for inflation to try to track prices to all the transactions level data. There are apps that people are using as you're just talking about, and the agencies are trying to figure out how do they inhale that fire hose of data. 2022 tax data it's too old that's not good enough and is there private sector data that we could use that would provide reliable statistics about what's happening right now? And I think that is a big open question.
Larry Bernstein:
You spoke to this group in 2017. In your discussion on business formation, you highlighted that most new businesses were in retail, small shops in the strip malls. You work with Eric Hurst who teaches at the University of Chicago’s Booth School, and his research suggests that most new small businesses do not want to grow. Instead, the entrepreneur wants to be his own boss and be in control of his hours and his destiny.
John Haltiwanger:
Eric Hurst and I are constantly debating the question you just asked, Larry, what do small businesses do? Eric's of the view that most businesses are small. 90% of employer businesses in the United States have less than 10 employees of employer businesses. Only 1% have more than 500 employees.
Can you identify ex ante that is early on is this going to be a business that takes off? It's incredibly hard to do that. Some industries we see much more skewed growth rate distributions. By that I mean much greater likelihood of right tail businesses, very fat rapid growth businesses. And high tech is one of those industries.
One view might be we should try to find ways of tracking the potential high growth businesses. Do I think that is a good idea? You betcha. Because it turns out they are the dominant players in terms of productivity and innovation.
But having said that, I am going to come back to my million jobs a quarter numbers show that I'm going to make it up in volume. So there's so many of them, they contribute literally millions of jobs. So even though these retail strip mall businesses are not the next big thing. They are a critical part of job creation in the United States.
What's the most likely outcome of a startup?
Larry Bernstein:
Death?
John Haltiwanger:
Death, what is the next most likely outcome?
Larry Bernstein:
Survival.
John Haltiwanger:
Survival, but no growth.
But there is a right tail of businesses that do phenomenally well. There is a right tail of high growth businesses literally in every sector of the economy.
Larry Bernstein:
There is a life and a death of a business. We start new businesses, it grows, it does something new and different. And then large firms acquire them, and that fundamentally changes their combined businesses.
After the merger, they downsize. The merger often results in firings when management decides who the most productive employees will be in the combined entity.
Those people who lose their jobs are gobbled up by small new upstarts and the process begins anew. Tell us about why business formation is important to the ongoing endogenous growth of our economy.
John Haltiwanger:
First let us talk about this changing organizational structure of the high-end businesses. The high-tech sector exhibits this like nobody else. In some respects, they have grown in terms of sales much more than employment and partly because they have not necessarily brought along all these support activities. Accounting, janitorial, staff security, you outsource all that. And so, an important phenomenon in the US is contracting out. I do not mean just contract out to individuals, contracting out to other businesses, all the support activities of businesses. And that is important in terms of rising earnings is the outsourcing of businesses from the top.
You asked, why is business formation important?
The radical innovations come from young businesses not mature businesses. At least one hypothesis out there, is that the large mature businesses are interested in protecting their market share and do not want to cannibalize what they have got.
When Google acquires one of these young startups with a new technology, are they acquiring it to innovate or are they trying to suppress it?
One reason we care about startups is innovation and productivity growth, but another reason is a critical role in job creation in the United States and oftentimes they are the entryway into millions of jobs. And when we see a dearth of these startups, which we did for much of the 21st century, the economy is quite anemic. We don't need all these business startups to be the next Google. They can be the strip mall businesses, but if they are fulfilling some new niche given the changing patterns of activity in the US economy, they are playing a critical role.
Larry Bernstein:
I want to call on Dave Bunning in the audience to tell us about his experience investing in biotech companies, and why new products are coming out of VC funded startups that are then acquired by big pharma. Dave, can you explain what you are seeing?
Dave Bunning:
Big firms like to have predictable earnings. The startup business is very unpredictable. If the innovation comes out of a university and it's got hope, there's still 20% chance it's ever successful. The big companies like buying the technology when it is proven and not taking the risk. The venture capital marketplace takes the $10 million trial, the $50 million trial, maybe the $100 million, but once it gets bigger than that it gets taken out by large pharma.
John Haltiwanger:
In other innovative intensive industries, back in the 1990s, the high-tech businesses wanted to be the next Google, and now increasingly that young high-tech businesses in that same space want to be acquired by Google. That is what you need because we now live in this networked, globalized environment.
Productivity growth surged in the United States from the late 1980s through the early 2000s during that tech boom and then it's been anemic since then. I do not want to be the next Google, I want to be acquired by Google is a good thing, why isn't it in the productivity statistics?
Larry Bernstein:
When I started on Wall Street in 1987, my friends worked for big investment banks like Salomon Brothers, Goldman Sachs, and Merrill Lynch. Today none of my friends work there, and nearly all of them have done something entrepreneurial in finance and started their own firms. This idea of entrepreneurship in finance was unknown when I started in the business, what happened?
John Haltiwanger:
Industries go through life cycles. There is a good reason many of you worked for Salomon Brothers or Goldman Sachs. They were incredibly successful firms in their history. They did something phenomenally well. Some business model, some business practices, some products, some services that were incredibly good in a healthy market economy. That is what you want. You want it to be the case that these right tail businesses, typically when we look at the data in terms of innovation or productivity, they are just out in the tail. Not in terms of growth, but in terms of productivity and innovation.
Amazon's a poster child for this. You want them to become incredibly large, but once they become large in a particular industry the question is are they up to the task of responding to the changing economic environment? And the evidence is mixed. Some large businesses that were historically dominant reinvent themselves, but not all of them.
IBM? Yes. Xerox? No.
There is changing technology, changing economic conditions, disruptive events, and the evidence seems to suggest that young businesses are especially good in these disruptive periods.
Jeff Bezos said 20 or 30 years from now, there is likely to be a new best thing.
Larry Bernstein:
The next question will come from my friend Larry Goodman. We used to work together back at Salomon Brothers and Larry is now the Founder and President of The Center for Financial Stability.
Larry Goodman:
There was an interesting piece that the IMF did bemoaning the drift lower in productivity and specifically citing creative destruction is not occurring as dramatically as it used to whether you look at total factor productivity or labor productivity. It peaked around 2000 and then drifted lower. When I listen to the conversation in the room, it sounds like creative destruction is alive and well. Why isn't this showing up in the productivity data?
John Haltiwanger:
I was here in 2017, I said exactly what you said, which is creative destruction was not so robust. If you look at the employer startup, it is much lower post-great recession than pre-great recession. And it's only now we recovered to the level of business formation. This is the employer businesses, that is the ones that matter for creative destruction. That is a long time.
Back in 2017, I was saying what the IMF was saying, creative destruction had slowed down. By the way, it is not just the startups that are the source of creative destruction. Exits is a fundamental part of a market economy. You want it to be the case that the businesses that are low productivity exit, that was happening to a much less extent as we move through the post-21st century.
In this most recent era, we have seen a surge in business formation, which is a canary in the mine for creative destruction. I am more optimistic that productivity growth is going to accelerate.
The entrepreneurs are the optimists out there think they see market opportunities that they just were not seeing back 10 or 15 years ago. I think creative destruction was low for a long time in the post-2000 period, but it's starting to rebound.
Larry Bernstein:
Thanks to John for joining us.
If you missed our previous podcast the topic was Putting Women Back to Work. Our speaker was Betsey Stevenson who is a Professor of Economics at Michigan. Betsey is the former Chief Economist at the Department of Labor. Betsey explained that the best way to maximize female employment is increasing labor flexibility. And discouraging the idea that the work product is necessary right now.
I would like to make a plug for our next podcast with Douglas Holtz-Eakin with the American Action Forum and is the former Director of the Congressional Budget Office. Douglas will discuss the need for the Federal Government to become more fiscally responsible.
You can find our previous episodes and transcripts on our website whathappensnextin6minutes.com. Please follow us on Apple Podcasts or Spotify. Thank you for joining us today, goodbye.
Check out our previous episode, Putting Women to Work, here.
Share this post