research on entrepreneurial finance: how to use the surveys of small business finances
DESCRIPTION
Research on Entrepreneurial Finance: How to Use the Surveys of Small Business Finances. Rebel A. Cole Departments of Finance and Real Estate DePaul University. 2010 Annual Meetings of the Financial Management Association October 21st, 2010. What is the SSBF?. - PowerPoint PPT PresentationTRANSCRIPT
Research onResearch onEntrepreneurial Finance:Entrepreneurial Finance:
How to Use the How to Use theSurveys of Small Business FinancesSurveys of Small Business Finances
Rebel A. ColeRebel A. ColeDepartments of Finance and Real EstateDepartments of Finance and Real Estate
DePaul UniversityDePaul University
2010 Annual Meetings of the Financial Management Association
October 21st, 2010
What is the SSBF?What is the SSBF?
Series of four nationally representative Series of four nationally representative surveys of U.S. small businesses.surveys of U.S. small businesses.
Small business: nonfarm, nonfinancial firm Small business: nonfarm, nonfinancial firm with fewer than 500 employees.with fewer than 500 employees.
Cross-sectional snapshots as of 1987, Cross-sectional snapshots as of 1987, 1993, 1998, 20031993, 1998, 2003
Not exclusively entrepreneurial firms, but Not exclusively entrepreneurial firms, but covers both entrepreneurial and non-covers both entrepreneurial and non-entrepreneurial firms.entrepreneurial firms.
Cole (2010)Cole (2010)Bank Credit, Trade Credit or No CreditBank Credit, Trade Credit or No Credit
What is the SSBF?What is the SSBF?
By looking at younger, faster growing By looking at younger, faster growing firms, you can segment the survey into firms, you can segment the survey into entrepreneurial and non-entrepreneurial entrepreneurial and non-entrepreneurial groups.groups.
Broadly representative of about 5 million Broadly representative of about 5 million predominantly privately held firms in each predominantly privately held firms in each year.year.
Very small number of publicly traded firms Very small number of publicly traded firms that can easily be excluded from the that can easily be excluded from the sample to target “privately held firms.”sample to target “privately held firms.”
Cole (2010)Cole (2010)Bank Credit, Trade Credit or No CreditBank Credit, Trade Credit or No Credit
What is in the SSBFs?What is in the SSBFs?
Data. LOTS of data.Data. LOTS of data. Thousands of variables.Thousands of variables. Major Sections:Major Sections:
• Governance/OwnershipGovernance/Ownership• Inventory of Financial ServicesInventory of Financial Services• Most Recent Loan ApplicationMost Recent Loan Application• Trade CreditTrade Credit• Balance Sheet and Income StatementBalance Sheet and Income Statement• Credit history of firm and ownerCredit history of firm and owner
Cole (2010)Cole (2010)Using SSBFs for ResearchUsing SSBFs for Research
What is in the SSBFs?What is in the SSBFs?
1987: 3,404 obs. / 3.2 million1987: 3,404 obs. / 3.2 million 1993: 4,637 obs. / 4.9 million1993: 4,637 obs. / 4.9 million 1998: 3,561 obs. / 5.1 million1998: 3,561 obs. / 5.1 million 2003: 4,240 obs. / 6.0 million2003: 4,240 obs. / 6.0 million
Cole (2010)Cole (2010)Using SSBFs for ResearchUsing SSBFs for Research
What is a sampling weight?What is a sampling weight?
Stratified random samplesStratified random samples• Stratified on Stratified on
Firm size: 0-19, 20-49, 50-99, 100-499Firm size: 0-19, 20-49, 50-99, 100-499 Urban/RuralUrban/Rural 9 census regions9 census regions
• Total of 72 sampling strata in 2003Total of 72 sampling strata in 2003 Each firm is associated with a sampling Each firm is associated with a sampling
weight.weight. It is important to use the sampling weights It is important to use the sampling weights
in your analysis.in your analysis.Cole (2010)Cole (2010)
Bank Credit, Trade Credit or No CreditBank Credit, Trade Credit or No Credit
What is a sampling weight?What is a sampling weight?
Sampling weight = inverse of selection Sampling weight = inverse of selection probability.probability.
1 out of 100 = 100 weight1 out of 100 = 100 weight 1 out of 1,000 = 1,000 weight1 out of 1,000 = 1,000 weight In other words, how many firms in the In other words, how many firms in the
sampling frame does a particular sampled sampling frame does a particular sampled firm represent?firm represent?
Large firm oversampled: 100 weightLarge firm oversampled: 100 weight Small firm not oversampled: 1,000 weightSmall firm not oversampled: 1,000 weight
Cole (2010)Cole (2010)Bank Credit, Trade Credit or No CreditBank Credit, Trade Credit or No Credit
What is imputation?What is imputation?
Imputation is the process of filling in Imputation is the process of filling in missing information.missing information.
Each firm is asked to answer all of the Each firm is asked to answer all of the survey questions, but some cannot or will survey questions, but some cannot or will not answer certain questions.not answer certain questions.
Without imputation, you would lose most Without imputation, you would lose most of your observations in multiple of your observations in multiple regression.regression.
Each variable has a shadow variable Each variable has a shadow variable indicating whether the value is reported or indicating whether the value is reported or imputed.imputed.
Cole (2010)Cole (2010)Bank Credit, Trade Credit or No CreditBank Credit, Trade Credit or No Credit
What is multiple imputation?What is multiple imputation?
2003 SSBF is a multiply imputed survey 2003 SSBF is a multiply imputed survey that provides 5 implicates.that provides 5 implicates.
Each implicate is identical for questions Each implicate is identical for questions answered by respondents.answered by respondents.
Each implicate is slightly different for Each implicate is slightly different for questions not answered by respondents.questions not answered by respondents.
This enables you to adjust for the This enables you to adjust for the variability introduced by the imputation variability introduced by the imputation process.process.
Cole (2010)Cole (2010)Bank Credit, Trade Credit or No CreditBank Credit, Trade Credit or No Credit
What is multiple imputation?What is multiple imputation?
In the 2003 SSBF, only about 1% of the In the 2003 SSBF, only about 1% of the responses were imputed so, for most responses were imputed so, for most variables, this is a non-issue.variables, this is a non-issue.
You can test the robustness of your results You can test the robustness of your results by rerunning your analysis with each of by rerunning your analysis with each of the five implicates to see if anything really the five implicates to see if anything really changes.changes.
If so, you might want to use survey If so, you might want to use survey procedures that adjust for this problem.procedures that adjust for this problem.
Cole (2010)Cole (2010)Bank Credit, Trade Credit or No CreditBank Credit, Trade Credit or No Credit
What is multiple imputation?What is multiple imputation?
DON’T run your analysis using all 21,200 DON’T run your analysis using all 21,200 observations in your analysis unless you observations in your analysis unless you are using MI software.are using MI software.
There are only 4,240 firm observations.There are only 4,240 firm observations. DO use only one of the five implicates.DO use only one of the five implicates. DO test for robustness by rerunning your DO test for robustness by rerunning your
analysis using each implicate.analysis using each implicate.
Cole (2010)Cole (2010)Bank Credit, Trade Credit or No CreditBank Credit, Trade Credit or No Credit
What about the earlier SSBF?What about the earlier SSBF?
Unfortunately, there is only one implicate Unfortunately, there is only one implicate for the 1987, 1993 and 1998 SSBFs.for the 1987, 1993 and 1998 SSBFs.
And there percentage of values that were And there percentage of values that were imputed is much higher than for 2003.imputed is much higher than for 2003.
Not much can be done about this problem.Not much can be done about this problem.
Cole (2010)Cole (2010)Bank Credit, Trade Credit or No CreditBank Credit, Trade Credit or No Credit
Can I pool the four SSBF?Can I pool the four SSBF?
Sure, you can, but I have no idea how to Sure, you can, but I have no idea how to interpret your results if you do.interpret your results if you do.
Samples are not independent. If you use Samples are not independent. If you use the sampling weights, you are double-the sampling weights, you are double-counting much of the sampling frame.counting much of the sampling frame.
The 5 million firms in the 2003 frame are The 5 million firms in the 2003 frame are largely the same as the 5 million firms in largely the same as the 5 million firms in the 1998 sampling frame, except for births the 1998 sampling frame, except for births or deaths.or deaths.
Good luck adjusting the weights!Good luck adjusting the weights!Cole (2010)Cole (2010)
Bank Credit, Trade Credit or No CreditBank Credit, Trade Credit or No Credit
Can I pool the four SSBFs?Can I pool the four SSBFs?
Also, beware nuanced changes in the Also, beware nuanced changes in the variables across time.variables across time.
Example: Most Recent LoanExample: Most Recent Loan• 1987, 1993: includes renewals of credit 1987, 1993: includes renewals of credit
lines but these are not identifiedlines but these are not identified• 1998: excludes renewals of credit lines1998: excludes renewals of credit lines• 2003: includes renewals of credit lines 2003: includes renewals of credit lines
and IDs new lines from renewals.and IDs new lines from renewals. Renewals are turned down at far lower rate.Renewals are turned down at far lower rate.
Cole (2010)Cole (2010)Bank Credit, Trade Credit or No CreditBank Credit, Trade Credit or No Credit
Best Thing about the SSBFsBest Thing about the SSBFs
They’re FREE!!!They’re FREE!!!
http://www.federalreserve.gov/pubs/oss/oss3/nssbftoc.htm
CodebooksCodebooks Methodology ReportsMethodology Reports Questionnaires, and . . . Questionnaires, and . . . DATA!!! DATA!!!
Cole (2010)Cole (2010)Using SSBFs for ResearchUsing SSBFs for Research
What can you do with the SSBFs?What can you do with the SSBFs?
Importance of lending relationshipsImportance of lending relationships• Petersen and Rajan (JF 1994, QJE 1995)Petersen and Rajan (JF 1994, QJE 1995)• Berger and Udell (JB 1995)Berger and Udell (JB 1995)• Cole (JBF 1998)Cole (JBF 1998)
Lending by large and small banks:Lending by large and small banks:• Cole, Golderg and White (JFQA 2004)Cole, Golderg and White (JFQA 2004)• Berger et al. (JFE 2005)Berger et al. (JFE 2005)
Cole (2010)Cole (2010)Using SSBFs for ResearchUsing SSBFs for Research
What can you do with the SSBFs?What can you do with the SSBFs?
Discrimination in SB credit marketDiscrimination in SB credit market• RaceRace
Cole (mimeo 1998)Cole (mimeo 1998) Cavalluzo and Cavalluzo (JMCB 1998)Cavalluzo and Cavalluzo (JMCB 1998) Cavalluzo et al. (JB 2002)Cavalluzo et al. (JB 2002) Blanchflower et al. (RESTAT 2003)Blanchflower et al. (RESTAT 2003) Blanchard, Zhou, Yinger (JUE 2008)Blanchard, Zhou, Yinger (JUE 2008)
• GenderGender Cole and Mehran (mimeo 2009)Cole and Mehran (mimeo 2009)
Cole (2010)Cole (2010)Using SSBFs for ResearchUsing SSBFs for Research
What can you do with the SSBFs?What can you do with the SSBFs?
Corporate GovernanceCorporate Governance• Agency costs and ownership structureAgency costs and ownership structure
Ang, Cole and Lin (JF 2000)Ang, Cole and Lin (JF 2000) Bitler et al. (JF 2005)Bitler et al. (JF 2005)
• Executive compensationExecutive compensation Cole and Mehran (mimeo 2009)Cole and Mehran (mimeo 2009)
Capital StructureCapital Structure• Cole (mimeo 2008)Cole (mimeo 2008)
Cole (2010)Cole (2010)Using SSBFs for ResearchUsing SSBFs for Research
What can you do with the SSBFs?What can you do with the SSBFs?
Trade CreditTrade Credit• Rajan and Petersen (RFS 1997)Rajan and Petersen (RFS 1997)• Cole (mimeo 2010)Cole (mimeo 2010)
Cole (2010)Cole (2010)Using SSBFs for ResearchUsing SSBFs for Research
Question?Question?
THE ENDTHE END
Cole (2010)Cole (2010)Using SSBFs for ResearchUsing SSBFs for Research