An Estimated Structural Model of Entrepreneurial Behavior

Transcription

1 An Estimated Structural Model of Entrepreneurial Behavior John Bailey Jones 1 Sangeeta Pratap 2 1 Federal Reserve Bank of Richmond 2 Department of Economics Hunter College and Graduate Center CUNY September, 2017 Disclaimer: The opinions and conclusions are solely those of the authors, and do not reflect the views of the Federal Reserve Bank of Richmond or the Federal Reserve System. Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

2 Motivation What drives entrepreneurial behavior? Financial constraints? Nonpecuniary benefits from operating business? Appetite for risk? Answer is key to policy design (Hurst and Pugsley, 2011, 2014) should we provide funding for start-ups? should we assist established businesses? Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

3 Our Approach Structural model of entrepreneurial behavior risk-averse entrepreneurs face borrowing + liquidity constraints limited liability and ability to exit to wage work encourage risk-taking nonpecuniary benefit (utility) to operating own business To our knowledge, first estimated model to consider all factors simultaneously: important for identification Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

4 Our Approach Structural model of entrepreneurial behavior risk-averse entrepreneurs face borrowing + liquidity constraints limited liability and ability to exit to wage work encourage risk-taking nonpecuniary benefit (utility) to operating own business To our knowledge, first estimated model to consider all factors simultaneously: important for identification Estimate this model using a unique dataset decade-long panel of New York state dairy farms detailed measures of both real and financial variables circumvents many econometric problems Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

5 Literature review Entrepreneurship literature decision to become an entrepreneur (Buera, 2009; Evans and Jovanovic, 1989; Hurst and Pugsley, 2011) wealth accumulation (Quadrini, 2000; Cagetti and De Nardi, 2006) distribution and composition of entrepreneurial wealth (Herranz et al., 2014) risk and limited liability (Kihlstrom and Laffont, 1979; Vereshchagina and Hopenhayn, 2009; Caggese, 2012; Caggese and Cuñat, 2013) Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

6 Literature review Entrepreneurship literature decision to become an entrepreneur (Buera, 2009; Evans and Jovanovic, 1989; Hurst and Pugsley, 2011) wealth accumulation (Quadrini, 2000; Cagetti and De Nardi, 2006) distribution and composition of entrepreneurial wealth (Herranz et al., 2014) risk and limited liability (Kihlstrom and Laffont, 1979; Vereshchagina and Hopenhayn, 2009; Caggese, 2012; Caggese and Cuñat, 2013) Structural corporate finance literature (Pratap and Rendon, 2003; Henessey and Whited, 2007; Strebulaev and Whited, 2012) Our paper complements: Caggesse s work with Italian data; Townsend and co. s work with Thai data Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

7 Results Preview Financial constraints are important determinants of real outcomes on the intensive and extensive margins collateral and liquidity constraints reduce capital and output significantly especially for large farms debt renegotiation allows productive farms to remain in business liquidation costs impede the exit of marginal farms Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

8 Results Preview Financial constraints are important determinants of real outcomes on the intensive and extensive margins collateral and liquidity constraints reduce capital and output significantly especially for large farms debt renegotiation allows productive farms to remain in business liquidation costs impede the exit of marginal farms Nonpecuniary benefits of farming play an important role many less productive farms would otherwise not be in operation interact with financial constraints Farmers are risk averse eliminating risk leads to larger operations but... expansion is small Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

9 Our Results in Context An understudied but important part of the firm distribution 78 percent of all US firms are owner-operated Revenues of average US firm $2.76 m ($2.89 m in sample) Sample distribution of firms similar to that of the US distribution Rare to find real and financial data for such firms Policy implications nonpecuniary benefits are important high-productivity farms are more financially constrained than low-productivity farms targeted financial assistance may work best Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

10 The Dairy Farm Business Summary Annual survey of New York State dairy farms conducted by Cornell University Unbalanced panel for farms, 2,222 farm-years Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

11 The Dairy Farm Business Summary Annual survey of New York State dairy farms conducted by Cornell University Unbalanced panel for farms, 2,222 farm-years Production (medians) Capital stock: $1.9 m 1/4 machinery, 1/2 real estate, 1/4 livestock. (186 cows) Revenues: $0.8 m Variable inputs (feed, fertilizer, small amount of hired labor): $0.6m Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

12 The Dairy Farm Business Summary Annual survey of New York State dairy farms conducted by Cornell University Unbalanced panel for farms, 2,222 farm-years Production (medians) Capital stock: $1.9 m 1/4 machinery, 1/2 real estate, 1/4 livestock. (186 cows) Revenues: $0.8 m Variable inputs (feed, fertilizer, small amount of hired labor): $0.6m Finances (medians) Assets: $1.9 m, Liabilities: $0.8 m All owner operated, 90 percent with 2 operators Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

13 Two Technologies Farms can either use stanchion barns or milking parlors Parlors larger, use more capital Allow for both technologies in our model Medians by Technology Stanchion Parlor No. of Farms No. of Operators 1 2 Total Capital 597 1,710 Herd Size (Cows) Output/Capital Intermediate Goods/Capital Investment/Capital Debt/Assets Cash/Assets Capital and herd size measured per operator Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

14 Measuring TFP - I Output/gross revenue for farm i at time t: Y it = z it M α i K γ it N it 1 α γ y it = Y it M i = z it k γ it n1 α γ it, Y it = output/revenue (in real dollars), M i = number of operators, K it = capital, N it = variable inputs (in real dollars) z it = total factor productivity (TFP) Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

15 Measuring TFP - II Output for farm i at time t: Estimation (below) yields: y it = Y it M i = z it k γ it n it 1 α γ, α = (0.107) and γ = (0.122) for stanchion (parlor) farms Decompose TFP as ln z it = µ i + t + ɛ it µ i = permanent farm-specific effect t = i.i.d aggregate shock ɛ it = i.i.d idiosyncratic shock Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

16 Aggregate Productivity t t is correlated with milk prices and cash flow. Significant financial risk 1.1 Aggregate Patterns in the DFBS: TFP Real Milk Price Cash Flow/Capital Year Aggregate TFP, Milk Prices and Cash Flow Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

17 Median Output by TFP Group: Data Median Capital Stock by TFP Group: Data Output (000s of 2011 dollars) Capital Stock (000s of 2011 dollars) Year Year Median Capital/Optimum Ratio by TFP Group: Data 0.5 Median Int. Goods/Capital Ratio by TFP Group: Data Capital/Optimum Ratio Int. Goods/Capital Ratio Year Year High µ farms have higher output, capital stock lower actual-to-optimal capital ratios higher intermediate goods to capital ratios Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

18 Median Debt/Asset Ratio by TFP Group: Data Median Dividends by TFP Group: Data Debt/Asset Ratio Dividends (000s of 2011 dollars) Year Year 600 Median Cashflow by TFP Group: Data 0.11 Median Gross Inv./Capital Ratio by TFP Group: Data Cashflow (000s of 2011 dollars) Gross Inv./Capital Ratio Year Year High µ farms have hold more debt, pay higher dividends have higher cash flow, invest more Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

19 Data: Conclusions Evidence that borrowing constraints matter high productivity farms do not immediately attain their optimal sizes aggregate cash flow and aggregate investment are positively correlated... even though aggregate shocks are serially uncorrelated Many low-productivity farms receive small dividend flows Our model reflects these considerations Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

20 Model Overview - I Families can either operate farms or be workers farming provides a non-pecuniary benefit (Hamilton, 2000; Moskowitz and Vissing-Jørgensen, 2002; Hurst and Pugsley, 2011) exit from farming is permanent exit from farming imposes liquidation costs Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

21 Model Overview - I Families can either operate farms or be workers farming provides a non-pecuniary benefit (Hamilton, 2000; Moskowitz and Vissing-Jørgensen, 2002; Hurst and Pugsley, 2011) exit from farming is permanent exit from farming imposes liquidation costs Limited liability for farmers may encourage risk-taking by marginal, highly-leveraged farms risk-taking incentive amplified by outside option of wage work (Vereshchagina and Hopenhayn, 2009) Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

22 Model Overview - II Risk-averse operators, know productivity type µ i learn aggregate shock t and idiosyncratic shock ɛ it after making production decisions Operating farms choose variable inputs, capital stock, liquidity (cash), and borrowing Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

23 Model Overview - II Risk-averse operators, know productivity type µ i learn aggregate shock t and idiosyncratic shock ɛ it after making production decisions Operating farms choose variable inputs, capital stock, liquidity (cash), and borrowing Working capital / liquidity constraints and collateral constraints Allow for debt renegotiation Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

24 Family Preferences - I E q Lifetime utility at age q: Q β h q [u(d h ) + χ 1{farm operating}] + β Q q+1 V Q+1 (a Q+1 ), h=q where d is the annual dividend per operator χ is the non-pecuniary benefit from operating a farm a Q+1 is the per-operator stock of assets at post-retirement age Q + 1 Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

25 Family Preferences - II E q Lifetime utility at age q: Q β h q [u(d h ) + χ 1{farm operating}] + β Q q+1 V Q+1 (a Q+1 ), h=q Functional forms with ν 0, c 1 c 0 0, θ 1. u(d) = 1 1 ν (c 0 + d) 1 ν, V Q+1 (a) = 1 (c 1 ν θ 1 + a ) 1 ν, θ Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

26 Worker s Problem V W q (a q ) = max d q u(d q ) + βv W q+1 (a q+1) s.t. a q+1 = (1 + r)a q + w d q, a q+1 0, where w is a constant age-independent wage. Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

27 Farmer s Problem: Timing Timing for a farm that operated at age q 1 Begin period q with outstanding liabilities b q and assets ã q : ã q = (1 δ)k q 1 + l q 1 + y q 1 n q 1 = (1 δ)k q 1 + l q 1 + z q 1 k γ q 1 n1 α γ q 1 n q 1, where l denotes liquid assets (cash) Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

28 Farmer s Problem: Timing Timing for a farm that operated at age q 1 Begin period q with outstanding liabilities b q and assets ã q : ã q = (1 δ)k q 1 + l q 1 + y q 1 n q 1 = (1 δ)k q 1 + l q 1 + z q 1 k γ q 1 n1 α γ q 1 n q 1, where l denotes liquid assets (cash) Choose between repaying debt in full and continuing renegotiating debt (b q replaced by ˆb q < b q ) and continuing liquidating and becoming a worker Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

29 Continuing Operations If the farm continues chooses d q, b q+1, k q, n q, l q time-t, age-q revenues are realized y q = z qt k γ q n 1 α γ q Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

30 Continuing Operations If the farm continues chooses d q, b q+1, k q, n q, l q time-t, age-q revenues are realized Sources of funds equity: e q = ã q b q (or ã q ˆb q ) y q = z qt k γ q n 1 α γ q debt (discount bond): b q+1 /(1 + r q+1 ) Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

31 Continuing Operations: Budget Constraints Sources of funds: equity e q and borrowing b q+1 /(1 + r q+1 ) Uses of funds: capital stock k q, dividends d q, liquidity l q e q + b q r q+1 = ã q b q + b q r q+1 = k q + d q + l q Financing investment: i t = k q (1 δ)k q 1 = (y q 1 n q 1 d q ) + (l q 1 l q ) + ( ) bq+1 b q 1 + r q+1 Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

32 Continuing Operations: Budget Constraints Sources of funds: equity e q and borrowing b q+1 /(1 + r q+1 ) Uses of funds: capital stock k q, dividends d q, liquidity l q e q + b q r q+1 = ã q b q + b q r q+1 = k q + d q + l q Financing investment: i t = k q (1 δ)k q 1 = (y q 1 n q 1 d q ) + (l q 1 l q ) + Financial constraints ( ) bq+1 b q 1 + r q+1 Liquidity constraint: n q ζl q, with ζ 1 Collateral constraint: ψb q+1 k q, with ψ 0 Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

33 Continuing Operations: Recursive Formulation V F q (e q, µ) = max u(d ( q) + χ + βe q Vq+1 (ã q+1, b q+1, µ) ) {d q c 0,b q+1 0,n q 0,k q 0,l q 0} V q+1 ( ) is the value of the farm before the occupational decision. Production risk: farmers must choose before shocks ɛ iq and t are realized Limits on dividends / new equity Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

34 Liquidation If the farm liquidates: Deadweight loss of λã q Farmers become workers with k q = 0 and assets a q a q = max {(1 λ)ã q b q, 0} Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

35 Renegotiation b q is replaced by ˆb q < b q ˆb q = max{(1 λ)ã q, b q} Lenders can force liquidation, bounding ˆb q from below Farmers can choose liquidation, bounding ˆb q from above V F q (ã q b q, µ) V W q (max{(1 λ)ã q b q, 0}) Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

36 Occupational Choice and Interest Rates Pre-choice value function: V q+1 (ã q+1, b q+1, µ) = max { V F q+1 (ã q+1 min{b q+1, ˆb q+1 }, µ), Interest Rates V W q+1 (max { (1 λ)ã q+1 b q+1, 0 } ) }. Renegotiation or liquidation decision determines the realized interest rate, r q+1 (ã q+1, b q+1, µ) The contractual interest rate r q+1 (ã q, b q, µ) is set so that where r is the risk free rate E q (1 + r q+1 ) = 1 + r Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

37 Estimation Strategy Calibrate w = $15,000 r = 0.04 δ (net of capital gains) = 0.02 λ = 0.35 Remaining parameters estimated through simulated minimum distance Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

38 Estimation Targets Split the sample along two dimensions Small vs. large (number of cows) Young vs. old Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

39 Estimation Targets Split the sample along two dimensions Small vs. large (number of cows) Young vs. old For each of these 4 groups, for each calendar year t, target median capital per operator, k t the median output-to-capital ratio, (y/k) t the median variable input-to-capital ratio, (n/k) t the median gross investment-to-capital ratio, (i/k) t the median debt-to-asset ratio, (b/ã) t the median cash-to-asset ratio, (l/ã) t the median dividend growth rate, d t /d t 1 Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

40 Estimation Targets Split the sample along two dimensions Small vs. large (number of cows) Young vs. old For each of these 4 groups, for each calendar year t, target median capital per operator, k t the median output-to-capital ratio, (y/k) t the median variable input-to-capital ratio, (n/k) t the median gross investment-to-capital ratio, (i/k) t the median debt-to-asset ratio, (b/ã) t the median cash-to-asset ratio, (l/ã) t the median dividend growth rate, d t /d t 1 Penalize counterfactual exit Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

41 Estimation Targets Let g mt, m {1, 2,..., M}, t {1, 2,..., T }, denote a summary statistic of type m in calendar year t, Model-predicted value of g mt is g mt (Ω). Our SMD criterion function is Q I (Ω) = M T m=1 t=1 ( g ) mt (Ω) Ψ I (Ω), g mt where Ψ(Ω) is a penalty imposed when farms exit in the simulations, but not in the DFBS. Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

42 Parameter Estimates Parameter Standard Parameter Description Estimate Error Discount factor β Risk aversion ν Consumption utility shifter (in $000s) c Retirement utility shifter (in $000s) c Retirement utility intensity θ Nonpecuniary value of farming χ C (consumption decrement in $000s) Returns to management: stanchions α Returns to management: parlors α Returns to capital: stanchions γ Returns to capital: parlors γ Strength of collateral constraint ψ Degree of liquidity constraint ζ Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

43 Model Fits - I Median Capital: Data (Solid) vs. Model (Dashed) Median N/K Ratio: Data (Solid) vs. Model (Dashed) Capital (000s of 2011 dollars) N/K Ratio Age Age Median Output/Capital Ratio: Data (Solid) vs. Model (Dashed) Output/Capital Ratio Age Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

44 Model Fits - I Median GI/K Ratio: Data (Solid) vs. Model (Dashed) Median Dividend Growth: Data (Solid) vs. Model (Dashed) GI/K Ratio Dividend Growth Age Age Median Debt/Asset Ratio: Data (Solid) vs. Model (Dashed) Median Cash/Asset Ratio: Data (Solid) vs. Model (Dashed) Debt/Asset Ratio Cash/Asset Ratio Age Age Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

45 Experiments: Nonpecuniary Benefits χ = 0: smaller farms exit, survivors have more assets and debt some farms in the DFBS have very low (and uncertain) incomes need a justification multiple interpretations, but Hurst and Pugsley s (2011) survey findings are very suggestive as is survey evidence for the UK & Europe (Fuchs-Schuldeln 2009, Binder and Coad 2013) Frac. Debt/ Cash/ Oper. Assets Debt Assets Assets Capital N/K I/K Baseline ,964 1, , χ = ,104 1, , Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

46 Liquidation Costs Eliminating bankruptcy costs (λ = 0) facilitates exit. similar to the effects of χ = 0 Liquidation costs and nonpecuniary benefits reinforce each other. substantially more exit, survivors have higher assets, debt, N/K optimal capital ( µ i ) much larger Liquidation costs and nonpecuniary benefits lead to financial inefficiency. Frac. Debt/ Cash/ Opti- Oper. Assets Assets Assets Capital N/K I/K mal K Baseline , , ,833 χ = , , ,070 λ = , , ,751 λ = χ = , , ,327 Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

47 Collateral Constraints Relaxing the constraint (ψ = 0.5) allows for more borrowing and more capital. raises value of farming, reduces exit investment rates fall by 2/3rds, at odds with data effects concentrated among larger farms Tighter constraint (ψ = 1.5 ) has opposite effect. capital stock falls by 27% more intensive use of intermediate goods, output drop is 19% Frac. Debt/ Cash/ Oper. Assets Debt Assets Assets Capital N/K I/K Baseline ,964 1, , ψ = ,405 1, , ψ = , , Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

48 Collateral Constraints (continued) 3500 Median Capital: Experiment (Solid) vs. Baseline (Dashed) Capital (000s of 2011 dollars) Age The Effects of Relaxing the Collateral Constraint (ψ = 0.5) on Capital Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

49 Collateral Constraints and Nonpecuniary Beneftis If nonpecuniary benefit (χ) is fixed at zero, estimated value of ψ falls to 0.76 if farmers receive no nonpecuniary benefits, they need higher profits easier credit to continue operating Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

50 Collateral Constraints and Nonpecuniary Beneftis If nonpecuniary benefit (χ) is fixed at zero, estimated value of ψ falls to 0.76 if farmers receive no nonpecuniary benefits, they need higher profits easier credit to continue operating If collateral constraint is eliminated (ψ = 0), estimated value of χ falls from 5.37 to 2.96 Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

51 Collateral Constraints and Nonpecuniary Beneftis If nonpecuniary benefit (χ) is fixed at zero, estimated value of ψ falls to 0.76 if farmers receive no nonpecuniary benefits, they need higher profits easier credit to continue operating If collateral constraint is eliminated (ψ = 0), estimated value of χ falls from 5.37 to 2.96 limited access to credit reduces profits farmers need more nonpecuniary benefits to stay in business. These mechanisms need to be considered simultaneously. Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

52 Liquidity Constraints Constraining input purchases to cash only (ζ = 1) has big effects changes composition of assets from capital to cash capital stock and output fall by 30% induces exit Looser constraint (ζ = 6) has opposite effects Frac. Debt/ Cash/ Oper. Assets Debt Assets Assets Capital N/K I/K Baseline ,964 1, , ζ = , , ζ = ,089 1, , Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

53 Risk Aversion Risk neutrality (ν = 0) No consumption smoothing motive maximize profits More aggressive investment and capital Lower debt, as β(1 + r) > 1 Internal financing, with negative initial dividend Matching observed (flat) dividend behavior a high value of ν Frac. Debt/ Cash/ % Divd. Oper. Assets Assets Assets Capital N/K I/K Growth Baseline , , ν = , , N.A. ν = , , Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

54 Risk Aversion (continued) Median Debt/Asset Ratio: Experiment (Solid) vs. Baseline (Dashed) Debt/Asset Ratio Age The Effects of Imposing Linear Utility on the Debt/Asset Ratio Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

55 Eliminating Risk Two transitory shocks: aggregate and idiosyncratic Eliminating the shocks has bigger, but still modest, effects More capital and assets Higher indebtedness Vereshchagina and Hopenhayn s (2009) arguments do not apply High risk aversion: ν = 4.1 Operators are patient: 0.9% discount rate Marginal operations tend to be less indebted Frac. Debt/ Cash/ Oper. Assets Debt Assets Assets Capital N/K I/K Baseline ,964 1, , No Agg ,988 1, , Shocks No Risk ,043 1, , Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

56 Year Year 2014 Farm Bill Previous farm bill: ineffective price support (Data from Schnepf, 2014) 2000 Milk Price farm bill: margin support (Data from Schnepf, 2014) Milk Margins Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

57 Margin Support Model margin support ($6.50 per cwt) by truncating tail of aggregate shock distribution Margin support modestly increases capital Premium ($0.29 per cwt) significantly reduces capital Premium levied per unit of output, treated as reduction in TFP Returns to scale are 1 α, α 0.12 optimal scale very sensitive to changes in expected TFP Debt/ Cash/ Opt. Capital Debt Assets Assets I/K K Baseline 1,596 1, ,833 Margin Support, No premium 1,623 1, ,832 Margin Supoort, Premium 1, ,457 Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

58 External Validity Distribution of revenues in DFBS resembles distribution in SUSB 0.60 Distribution of Firm Revenues Less than to to to 2.5 All US 2.5 to to 7.5 Our Sample 7.5 to to 15.0 More than 15 Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

59 External Validity Distribution of revenues in DFBS resembles distribution in SUSB 0.60 Distribution of Firm Revenues Less than to to to 2.5 All US 2.5 to to 7.5 Our Sample Our results comport with earlier findings nonpecuniary benefits are important: Fuchs-Schuldeln (2009), Hurst and Pugsley (2011), Binder and Coad (2013) collateral constraints are important, but less restrictive than earlier estimates: Evans and Jovanovic (1989), Buera (2009); Herranz et al. (2014) Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / to to 15.0 More than 15

60 Conclusions Unique dataset allows us to construct and estimate a unified model of entrepreneurship Integrate financial constraints, risk preferences and nonpecuniary benefits Model can account for time series and cross sectional variation in the data Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

61 Conclusions Unique dataset allows us to construct and estimate a unified model of entrepreneurship Integrate financial constraints, risk preferences and nonpecuniary benefits Model can account for time series and cross sectional variation in the data Findings Financial constraints are important, especially for large farms Nonpecuniary benefits are important Farmers are risk-averse, not risk-loving, but... eliminating risk has very modest effects Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45

62 Conclusions Unique dataset allows us to construct and estimate a unified model of entrepreneurship Integrate financial constraints, risk preferences and nonpecuniary benefits Model can account for time series and cross sectional variation in the data Findings Financial constraints are important, especially for large farms Nonpecuniary benefits are important Farmers are risk-averse, not risk-loving, but... eliminating risk has very modest effects Our data provides strong identification would be valuable to study other datasets Jones, Pratap (FRBR and CUNY) Entrepreneurial Behavior September / 45