62
Innovation-Driven Entrepreneurial Ecosystems: A New Agenda for Measurement, Policy and Action Scott Stern, MIT, CBS and NBER OECD Blue Sky III Forum September, 2016 Lab for Innovation Policy and Science

Stern - Innovation driven entrepreneurial ecosystems: A new agenda for measurement, policy and action

Embed Size (px)

Citation preview

PowerPoint Presentation

Innovation-Driven Entrepreneurial Ecosystems:A New Agenda for Measurement, Policy and ActionScott Stern, MIT, CBS and NBEROECD Blue Sky III Forum

September, 2016

Lab for Innovation Policy and Science

Enabling Shared Understanding Through Meaningful Metrics

Leveraging Digitization and Big Data to Develop and Test New Metrics

Assessing Innovation-Driven Entrepreneurial Ecosystems in Real-Time and at an Arbitrary Level of Granularity

Field of Dreams3

#MITREAP

3

Some Thinkers such as Richard Florida imply that regions are engaged in a war for talent and the creative class to be the NEXT Silicon Valley4

#MITREAP

4

With many Silicon Xs strategies being implemented around the world.5

#MITREAP

5

we dont want 50 Silicon Valleys; we want 50 different variations of Silicon Valley, all unique from each other and all focusing on different domains.Marc Andreesen, 20146

#MITREAP

6

How can we invest in innovation-driven entrepreneurial comparative advantage?#MITREAP6

7

A global program designed to help regions accelerate economic growth & social progress through innovation-driven entrepreneurship (IDE) built upon a regions unique history, capacity and comparative advantage. Partner regions form stakeholder teams and commit to a two-year learning engagement working with MIT faculty and the broader REAP community through a series of action-learning activities to assess, build and implement a custom regional strategy for enhancing IDE ecosystems. Mens et Manus28@MIT_REAPThe MIT Regional Entrepreneurial Acceleration Program (MIT REAP)?

8

MIT REAP Framework9@MIT_REAP

SystemStrategyStakeholders

Better graphic?

Slide structure for the 3 Ss: overview- key elements example

SystemIDE Framework (iCep & eCap +linkages etc accelerated by catalysts )Inputs vs outputs whole> sum of partsEngagement of all stakeholders! / shared measurement All needed for effective and exclusive change

StakeholderPentacleBackbone organizationCollective impactShared measurement

StrategyVersion of framework, highlighting catalystHow do you choose strategy for implementation? What is a regional strategy?How to make directed investment and action to leverage what you have and identify how to push out the frontier in such a way that will compliment effortsStrategy choice- choosing and winning must win battle + catalyst !

Emphasize role of choice - Josh Lerner not Bolavard of Broken Dreams but fixed dreams ?9

Cluster Based Comparative Advantage10@MIT_REAP

System

I-CapE-CapIDE EcosystemEconomic ImpactSocial ProgressFoundational InstitutionsCluster Based Comparative AdvantageMIT Framework

SystemIDE Framework (iCep & eCap +linkages + clusters + accelerated by catalysts )Inputs vs outputs whole> sum of partsEngagement of all stakeholders! / shared measurement All needed for effective and exclusive change

10

MIT FrameworkStrategy

Determine regional comparative advantage & how itwill accelerate IDE creation and growth in region

AcceleratorsPrizes & CompetitionsEarly-stageCapitalApproachesDiasporaNetworks &Immigration PolicySelect YourREAPStrategicInterventions32@MIT_REAP

11

Engaging regions all over the world

12@MIT_REAP

Quick introductions (1 min/ region) name, organization & role in REAP team12

Stakeholders are Key for Designing and Implementing Acceleration Strategies.IDE Ecosystem-led growth is different from traditional economic development approaches

Collaboration across key stakeholders is crucial for collective impact and acceleration at the ecosystem level

EntrepreneurRisk CapitalCorporateGovernmentUniversityInnovationEcosystemStakeholderModel

Shared Metrics and Evaluation Enable Stakeholder-Led Accleration!

ActionIDE Ecosystem Acceleration

STAKEHOLDER-LED ACTION REQUIRES SHARED UNDERSTANDING.

SHAReD UNDERSTANDING IS ENABLED BY METRICS.

But HOW CAN WE DEVELOP REAL-TIME ACCURATE AND GRANULAR METRICS THAT CAN (POSITIVELY) ENABLE THIS PROCESS?

consider the debate over the state of American Entrepreneurship.

Americas great challenge is to bring about a substantial increase in the numbers of highly successful new companies Nothing less than the future welfare of America and its citizens is at stake.

Litan, Robert E. Inventive Billion Dollar Firms: A Faster Way to Grow. SSRN Working Paper #1721608 (2010).

Thanks to important and insightful work by John Haltiwanger, Ron Jarmin, Javier Miranda and coauthors, we know that benefit of startups for economic growth spring from a handful of fast-growing, young companies.

For innovation and entrepreneurship to propel the economy, we need more startups capable of meteoric growth.

17

The problem is that it is very difficult, if not impossible, to know at the time of founding whether or not firms are likely to survive and/or grow. This is true even with venture-capital backed firms

Ian Hathaway and Robert Litan, 2014. Declining Business Dynamism: Its for Real. Brookings Institution.

Cant tell from start which firms have the potential for meteoric growth

18

While the quantity-based measures of entrepreneurship and dynamism are on the decline, funding and salience of high-growth businesses seems to be on the rise!

Quantity Based Measures

Decker, Haltiwanger, Jarmin, and Miranda (2013,2015)Haltiwanger, Jarmin, Kulick, and Miranda (2015)Hathaway and Litan (2014a, 2014b)

Funding of Venture-Backed Firms

Quantity-based measures (BDS) document a 30 year decline in entrepreneurship and business dynamism , with only a very modest leveling off (and uptick in high-tech)

Outcome-based measures (PWC/NVCA MoneyTree Report) indicate high-growth entrepreneurship is on the riseearly stage angel and VC funding on a significant upswing (unit just recently)Growing number of students joining startupsFear of bubble

The Business Dynamics Statistics (BDS) provides annual measures of business dynamics (such as job creation and destruction, establishment births and deaths, and firm startups and shutdowns) for the economy and aggregated by establishment and firm characteristics. The BDS is created from theLongitudinal Business Database(LBD), a confidential database available to qualified researchers through secureFederal Statistical Research Data Centers. The use of the LBD as its source data permits tracking establishments and firms over time.19

How can we measure high-potential entrepreneurship in a systematic and timely way?

How can we utilize real-time granular measurement of entrepreneurial quality to gain new insights and accelerate innovation-deriven entrepreneurial ecosystems?

21

REAP and ENTREPRENEURIAL QUALITY MEAUREMENT TEAM

Three Steps to Measuring Entrepreneurial Quality(at or near the time of founding)Business RegistrationsDigital Signatures of Growth PotentialSuccess outcomes can be mapped to initial digital signatures

Our ApproachA novel approach to the measurement of entrepreneurial quality, combining three interrelated insightsBusiness Registration as a Practical Requirement for Growth. Some form of business registration (incorporation, partnership, LLC, etc) is a practical and non-controversial requirement for ventures seeking to achieve a meaningful growth outcome. Public, comprehensive and comparable over time and place.

Business registration is a practical and non-controversial requirement for ventures seeking a meaningful growth outcome. Public, comprehensive and comparable over time and location.

25

Examples of Business Registration

Our ApproachA novel approach to the measurement of entrepreneurial quality, combining three interrelated insightsBusiness Registration as a Practical Requirement for Growth. Some form of business registration (incorporation, partnership, LLC, etc) is a practical and non-controversial requirement for ventures seeking to achieve a meaningful growth outcome. Public, comprehensive and comparable over time and place.Markers of Entrepreneurial Quality are Observable at or Near the Time of Business Registration. Firms with the potential and ambition for a meaningful growth outcome likely have different start-up characteristics including measures directly observable within business registration records, as well as publicly available measures that can be matched in a systematic manner .

A Tale of Two Bookstores

Quick example: Tale of two bookstores

Around time of their founding, difficult to tell that Harvard Bookstore and Amazon would have such different trajectories.

So, what data and methods can we develop to find the firms that have the potential to be amazons in the sea of mom and pop bookstores?

How can we identify and map growth potential from the start?

Quantity based measures treat them as equivalent

EQ measures ranks Amazon in top 1% of distribution at time of founding28

Patenting and Firm Growth: Helicos Biosciences CorporationDecember 5, 2003: Founded in Cambridge

May 24, 2004: First Patent Application

December, 2009: IPO as HLCS

Our ApproachA novel approach to the measurement of entrepreneurial quality, combining three interrelated insightsBusiness Registration as a Practical Requirement for Growth. Some form of business registration (incorporation, partnership, LLC, etc) is a practical and non-controversial requirement for ventures seeking to achieve a meaningful growth outcome. Public, comprehensive and comparable over time and place.Markers of Entrepreneurial Quality are Observable at or Near the Time of Business Registration. Firms with the potential and ambition for a meaningful growth outcome likely have different start-up characteristics including measures directly observable within business registration records (firm name) as well as publicly available measures that can be matched in a systematic manner (e.g., have they applied for a trademark or patent?).Meaningful growth outcomes can be observed with a lag, creating the potential for a mapping between growth and start-up characteristics. Rather than assume the relationship between start-up characteristics and entrepreneurial quality, investigate relative importance of different factors by developing a predictive model of growth based on start-up characteristics

Growth and Start-Up Characteristics

Regress growth on start-up characteristics:

gi,r,t+k is a binary growth outcome (IPO or high value acquisition, but could be others)Xi,r,t are start-up characteristicss is lag allowed for growth to be achieved (6 yrs)Divide sample into training sample (70% of obs where growth is observed), test sample (30% of obs where growth is observed, for validation), and prediction sample (where growth has not yet been observed but for which prediction is possible).

Our ApproachA novel approach to the measurement of entrepreneurial quality, combining three interrelated insightsBusiness Registration as a Practical Requirement for Growth. Some form of business registration (incorporation, partnership, LLC, etc) is a practical and non-controversial requirement for ventures seeking to achieve a meaningful growth outcome. Public, comprehensive and comparable over time and place.Markers of Entrepreneurial Quality are Observable at or Near the Time of Business Registration. Firms with the potential and ambition for a meaningful growth outcome likely have different start-up characteristics including measures directly observable within business registration records (firm name) as well as publicly available measures that can be matched in a systematic manner (e.g., have they applied for a trademark or patent?).Meaningful growth outcomes can be observed with a lag, creating the potential for a mapping between growth and start-up characteristics. Rather than assume the relationship between start-up characteristics and entrepreneurial quality, investigate relative importance of different factors by developing a predictive model of growth based on start-up characteristics

Entrepreneurial quality is the estimated probability of growth given startup characteristics.

How do the digital signatures of companies predict growth?(NB: Prediction NOT causal)Change in the Probability of GrowthHas Short Name248%Firm Named after Founder-70%Corporation (Not Partnership or LLC)405%Trademark in First Year501%Patent and No Delaware Registration3,534%No Patent and Delaware Registration4,470%Both Patent and Delaware Reg.19,640%Sectoral ControlsIncluded State ControlsIncluded

Guzman and Stern, 2016.

Reports core empirical relationship (based on logit regression)Examines how presence or absence of startup characteristic correlates with the probability of growth (conditioning on the presence or absence of other characteristics)Substantial changes in the predicted likelihood of a growth outcome are associated with characteristics observable from business registration filings (nowcasting) as well as characteristics observable with a lagCorporations 400% more likely to growFirms with short names 250%x more likely to growFirms named after founding 70% less likely to grow.Firms that register in DE and apply for a patent are 196x more likely to growCan use findings to construct, for every registered firm, its underlying probability of growth at time of foundingProbability of an average firm is low (1/3500)For firms with multiple positive startup characteristics, this probability jumps to 1/100Startup characteristics are digital signatures not causal drivers of growthDigital signatures of growth ambition and potentialRegistering in DE or filing for a patent will not guarantee that the business that does them will growBut, the firms that have historically engaged in these activities have been associated with skewed outcomes .33

A New View of the Skew

Out of sample tests show method works

To test the predictive capability of the model and evaluate the skewness of entrepreneurial growth potential, compare estimated probabilities of growth outcomes to realized ones (in a ten-fold cross validation in out of sample test

65-72% of realized growth events fall within top 5% of the models estimated entrepreneurial quality distribution51% of realized growth events fall within top 1% of the distribution of EQ10-80% in top 10% of the distributionModel highly predictive. EQ highly skewed.

Characteristics are highly predictive of a very rare outcomeShow skewed nature of growth outcomeIndicate that distribution of initial quality is itself skewed from the start(It is not that all firms start out equal and some happen to grow) [[growth potential is a key dimension of heterogeneity among newly founded firms]]]

34

Entrepreneurial Quality Statistics

New Population-Level Entrepreneurship IndicesEntrepreneurship Quality Index (EQI). Average estimated entrepreneurial quality within a group of start-ups:

Regional Entrepreneurship Cohort Quality Index (RECPI). Expected number of growth events within a regional start-up cohort:

Regional Ecosystem Acceleration Index (REAI). The ratio of realized vs. expected growth events in a region:

Attributes:Panel or cross-sectionalArbitrary level of granularityNot necessarily geographic in scope

EQI: The sum of the quality score for each individual firm/number of firms in a given region for given yearTo create an index of entrepreneurial quality for any group of firms (i.e., within a cohort or satisfying a particular condition), take the average quality within the groupAggregate of quality at the region-year basis by estimating average of Theta (I,r,t) within that region{I r,t} all firms in region r and year t. Nr,t = number of firms in that region year; Theta for firm I in region r at time tExcludes any location specific measures (do not assume startups from a given location are associated with a given level of quality)Excludes time-based effects (e.g., changes in available financing)Measure of the potential of a region given the intrinsic quality of firms at birth, which can then be affected by the impact of the ecosystem or shocks to the economy between formation and growth outcome.RECPIAverage quality of new firms multiplied by the number of new firms within a given cohort-region REAIThe ratio of realized to expected growth events in a region__________________________________________________36

Where is Silicon Valley? Science, 2015

Address-Level Visualization of Silicon Valley

The Quality of Entrepreneurship in Kendall Square

Nowcasting Entrepreneurial Quality in Kendall Square

Size of bubble = number of high-potential growth firmsColor of bubble = level of entrepreneurial quality

CIC and Cambridge coworking center (biggest dot

Dark colored dots: wetlab accelerators and biotech lab spacescorporate (IBM accelerator)40

The Quality of Entrepreneurship in Kendall Square

Nowcasting Entrepreneurial Quality in Kendall Square

Size of bubble = number of high-potential growth firmsColor of bubble = level of entrepreneurial quality

CIC and Cambridge coworking center (biggest dot

Dark colored dots: wetlab accelerators and biotech lab spacescorporate (IBM accelerator)41

The State of American Entrepreneurship

The State of American Entrepreneurship

Great variation across regions in Entrepreneurial Quality/growth potential43

RECPI / GDP:The State of American Entrepreneurship Over TimeRECPI / GDP shows a sharp raise in potential during the late 1990followed by a drop (but NOT a collapse) in 2001and more moderate increase after the Great Recession.Nowcasted Index tracks closely and documents boom since 2010Peak ofdot-com bubbleGreat RecessionGuzman and Stern (2016)

44

Silicon Valley RECPI:@pmarca view of the world

The View from Silicon Valley with Marc Andreesens head to the right45

The State of American EntrepreneurshipIn contrast to the secular and steady decline observed in the BDS, RECPI has followed a cyclical pattern that seems sensitive to the capital market environment and overall economic conditions. US RECPI increase over the 1990s, with a spike in 1999-2000, followed by a decline to a higher level than had existed in the 1980s. After a decline at the beginning of the Great Recession, strong growth in US RECPI since 2010. 2014 registers the third highest level of US RECPI (and highest in absolute value).

Potential Versus Realized Growth OutcomesThe highest potential cohort occurs in 2000

But the best performing cohort occurs in 1996

Guzman and Stern (2016)

48

Entrepreneurship and Economic GrowthPossible to examine the relationship between regional entrepreneurship metrics and subsequent economic growth

Without an exogenous shift in the supply of entrepreneurial quantity and/or quality, no casual interpretation

But useful for evaluating growth impact of policies that set the table for certain state of entrepreneurship

For each MSA in our sample, we estimate (in logs):

Levels of GDP, EQI, and Obs are averages of 2001-2003GDP Growth is to average GDP of 2012-2014

Entrepreneurial Quantity Has a Weak Postiive Correlation with 10-year Economic GrowthElasticity: .04 (p sum of partsEngagement of all stakeholders! / shared measurement All needed for effective and exclusive change

StakeholderPentacleBackbone organizationCollective impactShared measurement

StrategyVersion of framework, highlighting catalystHow do you choose strategy for implementation? What is a regional strategy?How to make directed investment and action to leverage what you have and identify how to push out the frontier in such a way that will compliment effortsStrategy choice- choosing and winning must win battle + catalyst !

Emphasize role of choice - Josh Lerner not Bolavard of Broken Dreams but fixed dreams ?60

Enabling Shared Understanding Through Meaningful Metrics

Leveraging Digitization and Big Data to Develop and Test New Metrics

Assessing Innovation-Driven Entrepreneurial Ecosystems in Real-Time and at an Arbitrary Level of Granularity

Thank [email protected]

Regression of GDP Growth at MSA Level. Dependent Variable: LN(GDP,2012-2014/GDP,2001-2003)

Largest CitiesAll Cities

(1)(2)(3)(4)(5)(6)

Ln(Avg(Obs,2001-2003))0.0384*-0.01090.0170-0.00884

(0.0154)(0.0221)(0.0294)(0.0130)

Ln(Avg(EQI,2001-2003))0.0803***0.0641**0.0684**0.0494*

(0.0191)(0.0218)(0.0252)(0.0201)

Ln(Avg(GDP,2001-2003))0.0509**0.02380.01020.0145

(0.0173)(0.0120)(0.0285)(0.0182)

Constant-0.2180.809***-0.288*0.4260.4460.496

(0.141)(0.165)(0.138)(0.266)(0.285)(0.257)

N6363636363150

R-sq0.0930.2400.1580.2780.2830.062

Robust standard errors in parentheses * p