The Ghost of Christmas Past: Small Business is the Best Job Creator is a Truism, but that doesn’t mean it’s True
This article, The Ghost of Christmas Past, is the first in our Small Business Holiday Trilogy. The Ghost of Christmas Past describes how and why the Small Business Truism came to be, and how Evil Forces successfully questioned its validity. Is small business our best new net job creator?
In subsequent issues over the Christmas and New Years holidays we shall send out The Ghosts of the Small Business Present and Future which carry the research on Small Business job creation to the present day and into the future.
Whether or not small business is the real job generator of the American economy is a question that probably doesn’t need an answer. It is a FACT. For most economic developers that small business is creates the most jobs is a fact not to be questioned. Everybody says it, from the President of the United States (known in journalistic circles as POTUS) to the taxi cab driver from whom you get your stock tips (the Curmudgeon drove a cab for two years in Boston area back in the 1960’s: he didn’t play the stock market, THEN); it simply must be true. Frankly in the Curmudgeon’s mind, there is simply nothing that can compete within economic development circles with the notion that small business is the way to create the most jobs and revitalize your community. Even the media and politicians believe it. Indeed, the Curmudgeon himself must admit – he believed it himself in his time when he was at the helm of an economic development agency.
But, the Curmudgeon being the Curmudgeon, is the truism true? Where did this shibboleth come from? Is there any data to back it up? Is the data that exists based on job creation in the 1690’s? Worst of all worlds – is it partially true and partially false and therefore misleading?
To help and guide the reader to answer these questions that for going on three decades has bedeviled academics, economists (are they two different things?), economic developers and practically everyone else, we offer for your Christmas time reading (who are we kidding?) a Lord of the Rings-style Trilogy presented by the Dickensian-like three ghosts of small business job creation. This article is the first in the Small Business Trilogy, The Ghost of Truism’s Past, and it purports to both describe how and why the Small Business Truism came to be, and how Evil Forces questioned its validity–almost from Day One. In subsequent issues over the Christmas and New Years holidays we shall send out The Ghosts of the Small Business Truism Present and Future which carry the research on Small Business job creation to the present day and into the Future.
The Ghost of Small Business Job Creation Past: David Birch
Research since the 1980’s has linked small business as the single most important element in overall annual NET job creation in the American economy. But, this finding is controversial and has been consistently questioned in no small measure because the Truism is near impossible to measure and easily falls victim to serious methodological and statistical deficiencies. Scrooges abounded and they were right as often as they were wrong–and in their wake we were all confused. The methodological issues were, and are, complicated and the Truism’s deficiencies were exaggerated by inadequate databases which previous to 2000, or so, were not up to the task at hand. This problem would plague the Truism until the turn of the century.
Well, we turn to David Birch, Job Creation in America: How our smallest companies put the most people to work, Free Press, New York. 1987, and with James Medoff, “Gazelles”, in Solomon and Levenson”s Labor, Employment Policy and Job Creation, Westview Press, Boulder, CO, 1994. To critique Dr. Birch we shall focus on a body of research centered around Steven J. Davis, John C. Haltiwanger, and Scott Schuh, especially, “Small Business and Job Creation: Dissecting the Myth and Reassessing the Facts”, Small Business Economics, Vol 8, 1996, pp. 297-315. In order to update and test both Birch and Davis, et al’s critique we will synopsized David Neumark (University of California, Irvine and California Public Policy Institute), Brandon Wall (Stanford University) and Junfu Zhang (Clark University and Institute for the Study of Labor (Bonn, Germany)) who, one dark and stormy night, apparently got together to write a research report which was published by NBER in December 2008. The review is entitled “Do Small Businesses Create More Jobs? New Evidence for the United States from the National Establishment Time Series“.
[A word about NBER (www.nber.org). It was founded in 1920, and called the National Bureau of Economic Research (hence NBER – who wouda guessed); it is a private, nonpartisan nonprofit and is headquartered in Cambridge, Massachusetts (the Curmudgeon probably dropped people off there from his cab, economists are lousy tippers). It considers itself, and many others also consider it, as the “nation’s leading nonprofit economic research organization”. Twenty-two Nobel Economics prize winners and thirteen past Chairs of the President’s (POTUS, remember) Council of Economic Advisers have published in it over the years. In any case, Working Papers are published on a weekly basis and they represent the author’s current thinking and research – at the time of publication]
Birch’s bottom line message was the book’s title, the smallest companies put the most people to work, and he was arguably the first to quantify and support that position. Birch’s research demonstrated that companies with fewer than 20 employees-create 98% of all net new jobs.
Our approach is to use, when possible, the actual words of the authors (with page citation provided usually); we will take occasional liberties with the original wording to facilitate readability, but we pledge to do our best to be faithful to the author’s message. Usually, but not always, when the Curmudgeon wishes to distort, confuse and blather, he will place his ravings within brackets . When in doubt, if it sounds halfway intelligent, it will be a paraphrase of the original authors, otherwise, it’s the Curmudgeon.
So let’s begin. Anyone that’s been involved in the economic development game longer than twenty-five minutes knows that the small business fixation started with David Birch who first published the idea in 1979 and then definitively proclaimed it in 1987 (Job Creation in America: How Our Smallest Companies Put the Most People to Work, Free Press, New York, 1987). Birch’s bottom line message was the book’s title, the smallest companies put the most people to work, and he was arguably the first to quantify and support that position. Birch’s earlier research demonstrated that “66% of all new jobs in the United States during the 1969-1976 period were created by firms with 20 or fewer employees, and 81.5% were created by firms with 100 or fewer employees.” (p.1) [The book jacket of the 1987 edition claims “that companies with fewer than 20 employees-create 98% of all net new jobs”].
This alleged importance of small business stood in marked contrast to the prevailing wisdom of the time which was typified in J.K. Galbraith’s New Industrial State. Galbraith indirectly argued, or at least implied, indeed Birch himself quotes Galbraith by saying that “this part [small business] of the economic system is not insignificant” … but Galbraith implies the reverse. Galbraith and others who think that big business is the American economy don’t seem to realize fully that the bubbly, yeasty, creative segment is the small businesses segment” (p. 8-9, Birch, 1987).
Further, in his 1987 book, Birch calls attention to the ongoing switch from manufacturing to the services sector which Birch believed to be the principal growth sector of the then current American economy. Whatever the sector his data and analysis observed and demonstrated that these growth firms were predominately very small. “As we look through our microscope, we see economic growth being dominated by a relatively few highly innovative firms, most of which have started small and grew by creating a whole new way of making or doing something. We call them ‘high-innovation firms’ (p.4, Birch).
This observation is one of the very earliest which called attention to what has since become a major, arguably paradigmatic, approach to economic growth and economic development: innovation, creativity and smallness. In the minds of many these forces became so intertwined as to become incorporated into our Truism. [To be sure innovation economics had been dealing with the economic theory underlying all this for more than two decades. But the discussion largely was in the bleachers of the economics profession and even as late as 1987 (1988-Buffalo conference was the real coming out party of Krugman, Romer and their mentors, Solow and Lucas) had not even penetrated mainstream economics. Birch’s position in 1987 impacted mainstream professional commentary and the academic literature much more than the early innovation economists had].But, we should also alert the reader that Birch continured writing after 1987. His 1987 work would be the foundation of our Truism but in 1994, Birch radically amended his own approach (Birch and Medoff cited above). In his “gazelles” article he questioned whether size of the firm was itself the critical dynamic. Instead, he suggested that most net jobs are created by a small number of high impact firms which he called gazelles [now you know where that came from]:
“These gazelles move between small and large quickly–at various times in either direction–and to classify them by their size is to miss their unique characteristics: great innovation and rapid job growth”. (p. 163, Birch & Medoff)
The “gazelles” article concluded that these gazelles (about 4% of all firms) created jobs at a rate far exceeding their proportion of total firms. As cited in a 2008 study by Acs, Parsons, and Tracy (High-Impact Firms: Gazelles Revisited) done for the SBA, (which we will describe in great detail in our Ghosts of Christmas Future issue) , Birch, Haggerty and Parsons (1995) “concluded that gazelles account for all new jobs in the whole economy.” (p.5, Acs, Parsons, and Tracy). Birch had moved on and size, as we all know, doesn’t really matter.
In essence, Birch himself by 1995 had left his Truism behind and had moved on.
[So there! We have answered a couple of our initial questions. It was Birch who gave birth to the truism and the data on which it was based was not 1690 but rather 1969-1986]
In his 1987 book, cited above, Birch was quoted as proving that 82% “of employment growth via expansion and contraction” in the 1981-1985 period were by firms with less than 20 employees, and that 88.1% “of overall employment growth” were created by these firms (as cited on p.1 of Neumark et al-so forget about the bookjacket). These later authors inject this last bit of precision, the Curmudgeon suspects, because from it’s outset Birch’s numbers and his methodology came under also immediate attack. A literature was soon to develop over the next fifteen years which questioned both the Truism and the methodology on which it was based. Birch, having moved on, did not participate in his own defense.
So how did we forge the Truism? Neumark et al observe (we are still on page one) that Birch’s central argument about the critical role of small business in job creation really dovetailed with “the U.S. government’s long tradition of supporting small business”. Pointing out that since the SBA was created in 1953 the Congress and the federal government consistently promoted and provided programmatic assistance to small business. “Birch’s findings fed into this thinking and quickly become conventional wisdom” (p.1) and ever since then Birch has been used by the federal government to justify a long string of loan programs, tax credits and incentives, favorable regulations, and support and assistance programs [such as a Curmudgeon fav, the Small Business Development (Counseling) Centers].
“Birch’s findings fed into this thinking and quickly become conventional wisdom” and ever since then Birch has been used by the federal government to justify a long string of loan programs, tax credits and incentives, favorable regulations, and support and assistance programs”
To this very day, the SBS’s Office of Advocacy “still trumpets Birch’s findings” (p.2) but, unfortunately for a quick and easy read on this topic, the Bureau of the Census entered the fray in an effort to prove or disprove the Truism that small firms created the most jobs. With its vast array of databases the Census Bureau added both sophistication and complexity into the equation of whether or not small business is indeed the greatest job creator in the American economy. The Truism had acquired a life of its own, and having been abandoned by its author, a new literature arose to critique, support, or disprove the Truism’s core principal that size of firm (small-less than 20 employees) was the key factor underlying net job growth in America.
Scrooge Questions the Ghost of Small Business Job Creation
Scrooge-like, bah humbug, criticism of Birch was soon in coming. A couple of years after Job Creation in America, (1990 to be exact), a fellow, believe it or not, named Charles Brown (and James Hamilton and James Medoff) in Employers Large and Small (Harvard University Press, 1990) attacked the quality of the jobs which small business created. Comparing the small business jobs to those created by large firms, Brown et al found the former (small business) created rather lousy jobs, with less pay, less job stability, poor fringe benefits, inferior working conditions, less job training and exhibited higher turnover rates (p.2). This was a serious concern and it remains to this day “an important” concern but given that these concerns affected real people in their real lives it would not be of significant professional concern . The quality of small business jobs, however, is NOT Neumark et al’s primary focus.
For Neumark et al, the core critique of Birch’s numbers and his methodology originated from two 1996 works by Steven J. Davis, John Haltiwanger, and Scott Schuh, “Small Business and Job Creation: Dissecting the Myth and Reassessing the Facts” (Small Business Economics, Vol 8, 1996).
[Before discussing specifics about Davis’s critique of Birch, the Curmudgeon would offer that Davis’s works do seem to possess a bit of apples to oranges aspect. The two studies use (1) overlapping, but different time periods, (2) different data sets, and (3) and Davis studies only one sector, manufacturing, which Neumark footnotes may well follow different growth patterns than other sectors and it is, after all, a sector which has been in serious job decline since the 1950’s. Neumark cites a number of other researchers who also have concerns with a critique based on manufacturing during this period of time. So as we understand Davis et al’s critique of Birch we should realize it too was under almost immediate attact by others who were defending Birch, who, if you remember, had moved onto hunting gazelles]
Davis’s analysis refutes much of Birch’s Truism, pointing out several important and potentially serious concerns principally with Birch’s database and methodology. It is these additional concerns which, if held to be substantially correct, would damage the Truism and the pivotal role of small business in job creation. The reader ought to keep in mind that Birch’s original study represented then and still remains the core defense of the small business job creation Truism to this day. If Birch’s study could be successfully questioned, we would be back to “square one”.
The first bottom line problem with Birch’s methodology was that he classifies small business firms into size classes (i.e. less than twenty, twenty to fifty, etc) and then “examines the variation in job growth across the size classes”[which means a firm’s jobs in one year is contrasted to the same firm’s jobs in a second year, etc]. To Davis et al this size classification scheme is flawed and leads to a “regression fallacy” which, Curmudegon’s words, leads to an overestimation in the number of jobs created by small firms. (p.2) Subsequent research, cited by Neumark, suggests this has proven not to be a major concern with Birch’s work (p.5-6) but Neumark et al themselves will conduct research to assess the extent to which the regression fallacy does or does not exist.[Not to keep you in suspense, they conclude there is a regression fallacy and they will design a methodology to correct for it].
[In essence Davis et al, are attacking the “quality” and integrity of Birch’s old-style Dunn & Bradstreet data, implying that D&B is error-prone and far too imprecise to support Birch’s far-reaching conclusions]
Other problems alleged to be part of the size classification approach to estimating small business job are that two types of firms “will accidentally fall into the smaller size category: firms that are not small but have just experienced a transitory negative shock to their employment …; and firms that are not small but are mistakenly classified as small due to random measurement [error from the D&B database]. Both types of firms allegedly will ‘grow’ fast from one year to the next only because [they will return to their former size] of regression to the mean (p.4).
[In essence Davis et al, are attacking the “quality” of the old-style D&B data, implying that it is error-prone and far too imprecise to be pushed to justify Birch’s far-reaching conclusions. The Curmudgeon can himself testify that his personal experience with D&B data in this time period, used as a mailing list only, almost always had error rates exceeding twenty-five per cent. Birch no doubt cleaned this up somewhat, but how much is somewhat? At the time of original publication, these weaknesses, which are fundamental, were discounted or ignored]
Using a different method of calculation, Davis et al believes removes the regression fallacy, Davis then analyzes firms in the manufacturing sector from the Longitudinal Research Database for the period 1973-1988. They conclude “that employment growth appears to have ‘no systematic relationship to average plant size'” (p.3). Size does not matter!
The second bottom-line deficiency with which Davis et al were concerned was the database used by Birch (a derivative of Dunn & Bradstreet data) and they alleged that Birch’s data “were not particularly suitable for studying job creation-destruction dynamics for two reasons. First, early D&B data suggested a much higher total employment number than alternative data sources such as those from the Census or the BLS (larger denominator). Secondly, the early D&B data did a poor job in terms of capturing new businesses (p.8) and establishing if they were true “stand alone” business independent of a larger firm. [In the Curmudgeon’s underpowered mind, and, to the extent they are valid, these are serious criticisms and do impact negatively on the original research of Birch. They call into question the validity of Birch’s findings, and therefore would invalidate our small business Truism apart from the regression fallacy]
Neumark et al decide to review Davis et al’s critique and test both Birch and Davis et al methodology and conclusions. To this end they develop with the Bureau of the Census a new database, the National Establishment Time Series (NETS) [which is also a sophisticated derivative of Dunn&Bradstreet data]. NTES includes ALL US business establishments since 1989 (although Neumark only uses data from 1992-2004). This database allows for the clear identification of branch or subsidiary firms, and therefore permits researchers to isolate “stand alone” firms in all sectors, not just manufacturing, a total of about 13.1 million firms annually (p.7)
The purpose of Neumark et al’s Working Paper is to test, using a new database, whether Birch or Davis is correct. Neumark et al are concerned with the numbers of jobs created by small business and the methodology by which those numbers are derived. [Neumark et al assert that the earlier cited deficiencies of D&B data have since been corrected and Neumark et al have taken great pains to validate the data and assess for its reliability. They feel they have corrected additional deficiencies associated with rounding, and in particular they have refined and enhanced the “capture” of births by new firms significantly (p.9) For specific details how they accomplished this feat, the reader should consult the article itself]
Still, after all these refinements, adjustments, and sensitivity to data deficiencies, Neumark et al “find the NETS database, like all others, is not without its flaws, and has both advantages and limitations (p.10), but Neumark does believe the deficiencies associated with the earlier Birch data sets have been remedied quite well and in particular, NTES, constructed with more recent D&B data, has corrected the NTES and Census/BLS gap in number of firms so that issue “is not a serious concern”, and the capture criticism has been made “largely invalid” (p. 10).
[I don’t know about you, the reader, but all this data and methodology stuff has got me bored. So, it’s time for a mandatory three minute reflection on an interesting (which precludes economics and databases) topic of your choosing]
Time Up, back to work. [So unprofessional!] Anyway, Neumark et al, having constructed this spectacular database, fraught with improvements and limitations, proceed to test whether Birch or Davis was right. At this point, probably none of us really care, but I guess we should, so how did the research turn out? Their research takes two approaches to determining whether small businesses create more jobs than larger firms:
Following the tradition established by Birch and Davis et al, we divide businesses into different size categories and examine whether there is a significant difference in net job creation rates across firm and establishment size categories. Second, we non-parametrically estimate the relationship between employment growth and firm or establishment size to avoid any potential biases or loss of information introduced by using arbitrary size boundaries (p.10). [In this report we will concentrate only on the first approach and simply report on the findings of the second]
In studying the growth derived from size categories, Neumark take into account Davis’s incessant whining about the regression fallacy created by relying only on a single base year as the basis for determination that a firm is small. Neumark minimizes this error by calculating an “average size” of the firm throughout the time period under observation. The net effect of this adjustment is that a firm will be counted as “small” only if it stays small for at least two consecutive years. For each qualifying firm they calculate gross job creation and gross job destruction and subtract one from the other (I forget which they subtract from) and get, of all things, net job creation (which is their key measure)[All this took three pages in the Working Paper].
Well Maybe Bah Humbug was too strong: The Findings
Using the base line figure alone as the basis for defining small firm and calculating growth rates, Neumark et all with raw data initially reach findings that conform pretty nicely with those of Birch. Firms with 0-20 employees average annual rate of job creation was 9%-10% and every other size classification recorded negative results [they lost jobs annually]. Firms with 1,000 or more workers had “rather large negative net job creation rates” (p.13) [So let’s stop while we are ahead].
Neumark et al, however, did not stop there and using their “average size definition” of small, Neumark et al arrived at an adjusted job creation rate [remember this is designed to minimize all the errors and regression fallacy we have so nauseatingly been describing, i.e. Davis’s critique]. Average size definition reduces the net job creation rate from 10.9% to 2.9% .[This is probably not good if you are a pre-1994 Birch supporter]. Birch’s extremely high negative rates for firms in the higher size classifications, derived from using only base year size definition, become “more muted” (p. 13) in Neumark. BUT, “Nevertheless, overall there is still a clear negative relationship between the net job creation rate and firm size”, although there is “some evidence of a higher net job creation rate in the largest firms” (50,000 or more) (p.13). [This appears a bit two-faced to the Curmudgeon and his summary presented below is regarded by the Curmudgeon as a gentle questioning of the thrust of Birch’s message and the wide-sweeping power of small business net job creation]
In essence, Neumark does “see evidence consistent with the regression fallacy” (p.14) which was the core of Davis’s critique. “In this sense, … the concern of Davis et al. over Birch’s methodology is well justified. However, our results [from using average size adjustment] suggest that Birch’s conclusion is still valid, though to a lesser extent … (p.14)
Now comes the really painful part of this methodological run around the mulberry bush. [The reader can, at his or her option, take a second three minute break at this point. The Curmudgeon did]. Neumark now turns his attention to Davis et al and reviews his methodology.
[So far we have been talking about “firms”; now we have got to talk about “establishments”. Stay with me now! [An establishment is a business unit at a SINGLE PHYSICAL LOCATION, i.e. one specific address. A firm like Walmart can obviously have thousands of establishments and these individual establishments can cut across any number of NAICs code classifications. David Birch uses “firms” as his unit of analysis, but Davis et al used “establishments”. The NETS database used by Neumark has “establishments” as its unit of analysis, BUT the NETS database has a field which permits the researcher to determine if the “establishment” is “stand alone” (i.e. roughly equivalent to a firm) or is a “branch” of a multi-establishment firm. Still with me now?]
[Usually, the findings of any research are specific to the methodology employed and different methodologies can easily yield different findings. There may be truth in them-there-damned statistics, but only so long as you stay within the methodology. Not surprisingly, different ideologies use different methodologies]
[If nothing else, the reader new to this research garbage is realizing how much details and definitions matter in statistical and methodological research. Journalists, for good reasons and bad, simply report findings and ignore these boring aspects and they leave readers with core and profound misunderstanding in regards to the findings, i.e. are egg yolks good or bad for your heart? How about fish oil? Usually, the findings of any research are specific to the definitions and methodology employed and different methodologies can easily yield different findings. There may be truth in them-there-damned statistics, but only so long as you stay within the methodology. Not surprisingly, different ideologies use different methodologies. The Truism is really a creature of methodologies and definitions]
Anyway, back to establishments and Neumark’s findings. Using establishments instead of firms, Neumark reaches findings inconsistent with Davis (again, assuming you care). The rates of net job creation by firms less than twenty employees are 2.6%, compared to Davis’s 2.9% –and the findings for higher size classifications are correspondingly close to that derived from Davis. Using establishments, however, positive net job growth by firms less than 500 employees (which do not correspond to Birch) is also supported. The net net of firms versus establishments is that use of establishments mutes net job growth and it contradicts Davis’s conclusions that small firms do NOT create the highest rates of net job creation. Got it? Davis was wrong too–at least in part.
[By the way, Scrooge wants to know how job creation by large and small firms is affected by their rates of job destruction. Small firms have high rates of gross job destruction, much more than larger firms which are much less volatile. Firms greater than twenty employees “have relatively low gross job creation and destruction rates” (p. 15). [If the reader values stability, very small firms are not necessarily the way to go]. Also, small firms may create a lot of jobs, but do they result in the destruction of a lot of jobs–not in the same firm or establishment, but in other firms or establishments and other size classifications (p. 15-16). This is what creative destruction is all about. Gazelles often kill other community’s jobs. Also, Scrooge wonders how the business cycle affects the job creation of each size classification, i.e. are certain phases of the business cycle more prone to small business or large business net job creation? These days we need to know how recessions in particular affect this job creating festivity. And do sectors like manufacturing (mature industries), high tech (immature industries) and the now dominant service sector respond equally or differentially to job creation by different size firms?]
Neumark et al, it must be confessed, are generous to Birch and the Truism in that they are reluctant to make a judgment as to whether the muted levels of annual net job creation by firms under twenty employees has been reduced to 2.9% from Birch’s finding that 66% of all new jobs created are by such firms. Instead, they correctly observe, “there appears to be a robust finding that small businesses do create more jobs” (p. 22) but they do also concede that “the base-year size definition used by Birch badly overestimates net job creation rates for small firms or establishment” (p. 22). Neumark also has meaningful negative observations for Davis by stating that “the relationship between establishment size and job creation is negative for the manufacturing sector” while also confirming that Davis’s fundamental point, that the regression fallacy existed with crippling effects on Birch’s thesis and findings.
So both Birch and Davis were to some degree incorrect, if not wrong. And to some degree they were both right. What is Scrooge to make of that? Certainly, methodology matters in estimating net job creation and which size firms are the better engine of job creation. Secondly, yes, there is a fallacy of regression inherent in Birch and that is very damaging. Thirdly, Birch’s gazelle growth rates in firms less than twenty are reduced by 75% to somewhere less than 3% and more than 2.5% annually–far from barn burners, but, still, small, Tiny Tim, businesses do disproportionately create more jobs than large firms. They are large net job creators. Fourthly, large firms greater than 1,000 and establishments up to 500 employees can create positive net new jobs. We can’t ignore them. Firms of all sizes above twenty employees create jobs, just not as many as all the small, Tiny Tim firms combined.
Birch is not exactly refuted, but his great generators of job growth are reduced to a kind of bah humbug pace. Everybody was right and wrong — and the Truism, if you think about it, is the big loser. It remains true, but to what extent, and why, and who exactly.
Birch is not exactly refuted, but his great generators of job growth are reduced to a kind of bah humbug pace. Everybody was right and wrong — and the Truism, if you think about it, is the big loser. It remains true, but to what extent, and why, and who exactly. No community is going to turn around very quickly with that kind of job growth created by firms less than twenty employees.
It turns out that the Truism was constructed with a rather unsubstantial foundation which can support only a fairly modest structure. Size of firm, in itself, is not what creates new net jobs in sufficient quantity to revitalize or drive an economy. So does the Truism collapse under the weight of too grandoise expectations? Unlikely, there are still too many jobs involved, too many entrepreneurs, too much symbolism and metaphor, and a self-perpetuating professional industry built around supporting and enhancing small business well-being. The Truism is not going anywhere — but should economic developers look elsewhere if they want new net job creation?