The OECD has a new version of their timely indicators on entrepreneurship available today at the national level for 10 countries.  



These data show that new business formation continues to operate in most countries at the new levels which were reached after the 2008 recession (Australia stands out for its continued positive performance).  These are the most timely indicators available internationally and seem to show anemic start-up rates continue in most countries although the rates haven't worsened for the most part.  

This release is timed around the start of Global Entrepreneurship Week.  I would just note that there is much more data available from the OECD on more complicated and nuanced views of entrepreneurship.  Most of that data is current through 2007 or 2008.  

The Washington Post has a nice piece online today examining some of the claims and counter-claims about how small business owners would be impacted by tax increases on those earning more than a million dollars.  The Post is correct in its suggestion that the more narrow interpretation of the data is the better one.  Indeed, it's not surprising that if one is looking at millionaires that the vast majority have some ownership in a business of some sort.  When you have that much money you have to park it somewhere to make a return and owning a business seems a natural place (in the given data it's impossible to make attributions about how the millionaires made their money as none of the analysis is really longitudinal).  If you aren't looking at where they are also investing their time in running a business then you would be creating a group of people who is dominated by investors and not business operators.  While this might be something which is helpful in talking about angel investors, it's probably not a good characterization of the question being posed.

One piece we released last year is relevant to this discussion but has not been picked up by many in the mainstream discussion.  Business Owners, Financial Risk, and Wealth uses the Survey of Consumer Finance to examine how business owners appears similar and different to other groups in their wealth and risk preferences.  It shows that in most ways business owners (defined even more conservatively than the Post article suggests) are conservative in many of their borrowing and savings patterns but more likely to assume risk for return.  The study concludes: "The results suggest that policies aimed at increasing business ownership should focus on helping households identify high-value business opportunities through transparent tax, legal, and regulatory systems. Efforts to reduce risk should focus on the business venture, such as full loss offsets, rather than focusing on reductions in other financial risks." Worth a read for those following this debate.  

The Numbers Guy, Carl Bialik, over at the Wall Street Journal has a piece in today's edition looking at some of the claims about small business job creation occurring in the political debate.  It is an important topic, and I was glad to see him tackle it.  That said, I was disappointed in his treatment of the question.  I have a lot of respect for his columns and how he brings focus to important elements of the discussion that involve numbers.  With his treatment of small businesses and jobs, he talked to most of the people I would have talked to but ultimately didn't get into any of the depth that I would have expected.  Specifically, I don't think he looked at several important elements closely enough for readers:

• Small vs. new

I've written about job statistics over the years and most recently focused on them in our Starting Smaller; Staying Smaller piece.  And while I haven't jumped directly into the debate about small businesses job creation vs. new business job creation, that seems to be what Carl was starting to do but didn't really. Most of the job creation (and a lot of the destruction) seen when looking at questions about small business job creation comes from small, new businesses.  These are businesses which are both very small and also happen to be young but which statisticians historically have not been able to disentagle because the data wasn't stored to look at firm dynamics over time.  A lot of what we have worked on these past several years has been trying to bring this element of firm age into the discussion about how jobs are created and destroyed.  And while the column today talks some about this issue near the conclusion, I wasn't impressed with it's treatment of the topic.  I recommend the piece by Haltiwanger, Jarmin, and Mirinda on this topic.  When looking at jobs every lens you observe through has advantages and disadvantages.  I wouldn't claim looking through the added element of firm age is perfect, but to me it's still more helpful than just looking at firms smaller than 500 employees.  

• Local vs. non-local

The other piece I would have liked him to get into a bit which he did not is the lens used by Youreconomy.org.  This is a site run by the Edward Lowe Foundation that uses privately-sourced data, NETS, which is based on D&B records.  What Lowe does which I like and I think helps the debate so much is to look at "resident" and "non-resident" companies and job creation.  Essentially this just means that they are able to look at jobs which are created by companies headquartered in a particular region vs. those which are not.  They add to this a discussion of business size.  While I don't think this is the perfect lens I do think it is at least a helpful one for most local communities in understanding job creation.  Because big employers, Fortune 500 companies, for example, can be good employment contributors in some cases but the view you take on their employment might be quite different if they are headquartered in your region vs. if they are not.  

In Starting Smaller; Staying Smaller: America's Slow Leak in Job Creation, we focus on declining contributions of new and young firms to employment in the U.S. over the last decade.  And when people ask why this matters I say because "what you are born with is what you have to live with."  This is the most important thing I have learned from studying business dynamics at the national level.

What do I mean?  Look to the following chart from the report:



When we follow businesses born in the same year over time it becomes apparent that they follow some steady trends.  Namely, your total cohort employment tends to be fairly sticky at age 2 (the solid red line above), about as many hirings as firings in total, but by age 5 the total employment of the cohort has dropped to 90 percent of its initial total (dotted red line above).  Anything below 1 on this graphic means that the cohort of businesses has lost jobs in comparison to what they were born with; above it means the cohort has gained jobs in total.  Notice that the lines on this graphic are almost always below 1.  So, when a cohort of new businesses starts with less employment, it keeps less employment as it ages.  The BLS and Census data show that in the last 35 years, what a business cohort has started with in terms of employment is likely to be that cohort's maximum employment over time.  

As we point out in the report, the cohort of businesses started in 2009 began with between 700,000 and 1,000,000 employees fewer than would have been expected historically.  There is also evidence from BLS that this slow down in employment generation has been going on for much longer.  Thus, it seems we have been accumulating recent new business cohorts with less employment potential.  And indeed, what people fail to realize is that the population of U.S. businesses is just like the population of U.S. - a steady accumulation of years of small changes which often go unnoticed at any point in time (see Neutralism and Entrepreneurship for a good discussion).  In the U.S. case, new businesses provide a new lifeblood of business activity, fueling hiring, entry of more productive business concepts, and the like.  Individual U.S. businesses come and go but on the average there is a steady and slow accumulation of businesses, driven by new entrants over time.  With fewer and fewer jobs at the start and declining rates of employment retention in cohorts of businesses, as shown above, America's slow leak in job creation accumulates into a major part of America's jobs crisis.  

Anyone following my blog knows that I have been spending a lot of time looking at jobs numbers in the U.S. as of late and am happy to say today that my report is finally out.  Starting Smaller; Staying Smaller: America's Slow Leak in Job Creation is a look at long-term trends around job creation by young, U.S. employer firms.  Employer firms are important to track because they are the bigger starts, those with employees, and thus tend to have more of an employment impact on the larger U.S. economy.  I think for too long we have used tallies of new business starts as benchmarks of the health of the U.S. entrepreneurial system.  What this report shows is that young businesses have been undergoing some major change in the last decade, particularly related to their employment patterns, and that the immediate as well as accumulated effect of these changes is a major reason for the United State's current unemployment problem.  New businesses remain a critically important source for net job growth, but they are starting smaller and growing less in their first five years in comparison to historical trends. Throughout the week I'll be posting some portions of the report as well as some things that didn't quite make the final report but that I think are still very telling and of interest to a more technical audience.  Here are a few of the most important charts from the report.

The Bureau of Labor Statistics continues to improve their presentation of information relevant to the study of entrepreneurship.  "Entrepreneurship and the U.S. Economy" is a must read for policy-makers as it presents some very important aggregate trends in an up-to-date and simple to understand way.  That said, just once caution, as I am currently writing and researching using these data.  Some of the jobs trends seen in the BLS statistics and Census Bureau statistics are quite different and need to be looked at together.

A particular chart that stood out to me in BLS's publication is the following which really charts how the most recent recession most significantly impacted large and small businesses and medium-sized businesses weathered somewhat better:

Hats off to Andrew Reamer at Brookings for another great advocacy piece on the U.S. federal statistical system.  "Putting America to Work: The Essential Role of Federal Labor Market Statistics" is a call-to-action for policy makers to create a labor market statistical system which actually tracks labor needs, educational attainment, and other supply and demand issues in a more comprehensive and real-time manner.  It is time that we stopped obscuring the inequalities and the failures (and successes) of our education system through a hamstrung statistical system.  For too long we have operated blindly with regard to our labor inputs and outputs and Andrew has many concrete recommendations for bringing these measures together.  I wanted to call my readers' attention to a section where Andrew talks about some innovations occurring currently:

Local Employment Dynamics: LED links and analyzes millions of workforce administrative records, particularly establishment and employee wage records from state unemployment insurance systems. An experimental program for a decade, Congress approved permanent status for the program in 2009. With annual funding of $14 million, LED is in the first of a three-year expansion plan. LED’s Quarterly Workforce Indicators analyzes workforce dynamics such as hires, fires, turnover, and wage levels by geography (state, metro, county, workforce investment board) and demographic characteristics (age, sex). OnTheMap visualizes the relation- ship between where people work and reside. A third product, a job-to-job flows tool, will show how defined groups of workers (e.g., in a particular industry and geography, with particular demographic characteristics) move through the economy over time. LED is close to having 50-state coverage. Over the next two years, with proper funding, it plans to add worker characteristics of occupation, educational attainment, race, and ethnicity.

Statewide Longitudinal Data Systems: Statewide longitudinal data systems are NCES-supported, state-managed efforts that track individual progress through formal education programs (pre-kindergarten to postsecondary) and into the workforce. Forty- one states and DC have received NCES grants, totaling $515 million since FY2006. NCES and ETA are encouraging the matching of worker job, wage, and training history with academic history. The Census Bureau’s LED offers the potential to link education data with workforce outcomes across state lines. A complete set of 51 SLDS that link education and workforce microdata would greatly aid understanding of educational program outcomes and career path patterns and so inform student choices of careers and schools, employer choice of workers, educational program design, and public policies.

Real-Time LMI: Using $4 million in Recovery Act funds, ETA is supporting an eight-state consortium’s development of an innova- tive real-time LMI project for green jobs in the Northeast. Through use of intelligent software, job ads on the web are regularly “scraped” and analyzed to collect current and trend information about job vacancies by geography, occupation, industry, required levels of education and experience, and earnings levels. The information is auto-coded (into standard occupational and industry coding structures) and parsed (to categorize and understand the meaning of the words/phrases contained in the ads). The tech- nology eliminates the time lag between data collection and data production common to most publicly-produced data sources. Real-time LMI will enable vacancy rate estimation, six- and 12-month projections of occupational demand, and better under- standing of the demand for and supply of community college certificates and industry certifications. If successful, and with proper funding, real-time LMI technology could be applied to all regions and occupations.

Strategic Objective 3.3: Promote the availability, analysis, and dissemination of the most current, accurate, and detailed statistics possible on small business.

Strategies
1. Advocate for improved data collection on small business activity. Pursue new avenues for improved and expanded data products on small business by working with other government agencies and external sources.
2. Carryout and publish data research and analysis. Through both internal analysis and contract research, publish regular, useful, high-quality data and indicators on small businesses and the role that they play in the economy. 

3. Raise awareness of data and findings. Publicize the availability of data and findings to federal agencies, Congress, small business organizations, research organizations, the media, and other stakeholders.

Last fall I participated in a really unique workshop at Yale put on by one of our Kauffman legal fellows, Victoria Stodden.  The purpose of the workshop was to discuss data and code sharing best practices and issues for creating replicable research.  While the workshop was a bit more in the computational science space than I am fully comfortable, I found the conversation incredible and the goals of the effort beyond compelling.  What has resulted is a Data and Code Sharing Declaration (just published in IEEE Computing in Science and Engineering).  This is a document that should be taken up for discussion at foundation and other funder events, in policy circles, and within the scientific academia as it lays out early and clear recommendations for actions each group can take to further data sharing and replication in the future.  It is a document which anyone who curates data, journal editors, and all scientists should be discussing.  

A new paper out from the National Bureau of Economic Research - “Dynamic Text-Based Industry Classifications and Endogenous Product Differentiation” by Gordon M. Phillips and Gerard Hoberg - discusses the power large-scale text analysis can provide in examining industrial classifications and other traditionally nebulous areas of differentiation among firms and markets. 


This is a timely publication as the Office of Management and Budget (OMB) is in the final stages of seeking approval (and feedback) about the 2012 revisions to the North American Industrial Classification System.  While there is a lot of effort made to update these industry classifications unfortunately I do not believe that government officials are yet taking advantage of some of the methods which are described in this paper which mine existing data to look for discontinuities in how industries are defined, when firms change industries, or other aspects of industrial organization. 

Now, the prospect of the government performing large-scale text analysis like this might scare some, but in my mind, there are groups like the Center for Economic Studies at the Census Bureau or other places like the Statistics of Income Division at the Internal Revenue Service who could do this responsibly if given the mandate, funding, and some lead time.  These places house large quantities of text data yet maintain separate research functions and most importantly they maintain processes for seeking outside researcher proposals for cutting edge research which would benefit the agencies through improved data products.  I’ve never heard staff at either of these locations discuss this NAICS redesign as a high priority but perhaps if OMB were using their coordinating powers and discretionary funding with more force, that could change. 

Identifying new industries clusters, and other big changes in the industrial organization faster and more accurately remains a key deficiency in the current national statistical system.  The U.S. regions who are on the front line of economic development rely too much on private data to try to understand change in their economies because the federal system has too often missed the data needs of the diffused customers here.  Coincidentally, the Council for Community and Economic Research annual conference starts today in Washington, DC.  This is the most organized group of individuals advocating for improved regional economic statistics in the United States. 

I should note that while there is great potential power in the methods employed by Phillips and Hoberg, the authors also note the potential gaming which could be used by firms if they felt the text they were sharing could be manipulated to effect government policies to the firms advantage.  “We also note that
while our new measures are interesting for research or scientific purposes, they would not be good for policy and antitrust purposes as they could be manipulated by firms fairly easily if firms believed they were being used by policy makers.”  I think these methods would be best added to an existing review process and not seen as a substitute.  In that case, the ability to game the system could be reduced.