1. The Financing of R&D Part 1 Bronwyn H. Hall UC Berkeley and U of Maastricht

2. Outline Economics of R&D as investment Implications for financing R&D Asymmetric information Asymmetric information Agency costs Agency costsEvidence Implications for technology policy (limited)

3. Economics of R&D and innovation Arrow 1962 – market fails to allocate adequate resources to innovation because.. 1. Lack of full appropriability of returns unpriced positive externalities 2. Indivisibility of output implies market power for innovator from returns to scale 3. Financing is costly because of info asymmetry and risk especially when financier and entrepreneur are different people

4. R&D as investment Similarity: Expenditure undertaken today to secure (uncertain) returns in the future Expenditure undertaken today to secure (uncertain) returns in the futureDifferences: Composition – wages of scientists and engineers are more than half of spending Composition – wages of scientists and engineers are more than half of spending Asset created is intangible Asset created is intangible Unknown share is human capital (partly owned by employees) Not easily tradeable (low salvage value) Level of uncertainty much more extreme Level of uncertainty much more extreme

5. Implications for policy and practice Production of knowledge is not intemporally separable → adjustment costs high Policy changes take time to have an impact Policy changes take time to have an impact Measurement difficulties - R&D does not exhibit much variation over time within a firm Measurement difficulties - R&D does not exhibit much variation over time within a firm Responds slowly to changes in capital cost (LHS) Little variation to identify its productivity (RHS) Uncertainty – in some cases, distribution of returns is Pareto (and without a second moment) risk adjustment problematic risk adjustment problematic

6. Modeling R&D investment Marginal profit condition: marginal product of R&D capital equals tax-adjusted user cost of capital: r = investor’s required rate of return δ = (economic) depreciation rate MAC = marginal adjustment cost τ = corporate tax rate A d = PDV of depreciation allowances A t = PDV of tax credits Last equality holds if R&D expensed and no special tax credit.

7. Implications for R&D finance Debt versus equity finance debt prefers physical assets as collateral but.. debt prefers physical assets as collateral but.. R&D creates an intangible asset R&D creates an intangible asset Tax treatment Tax subsidies to R&D in some countries but.. Tax subsidies to R&D in some countries but.. Debt sometimes cheaper than equity Debt sometimes cheaper than equity Depreciation (private obsolescence) highly variable and endogenous to other firms’ behaviors

8. Required rate of return to R&D Probably higher than that for ordinary investment: Uncertainty and risk Uncertainty and risk Asymmetric information between financier and firm implies there is a lemons premium Asymmetric information between financier and firm implies there is a lemons premium Mitigating asym info by revealing idea to potential investor is costly and can lead to imitation Akerlof (1970): if lemons premium large enough, market disappears One solution: hands-on venture capital investment Agency costs – can arise in any setting where the goals of a principal and his/her agent conflict Agency costs – can arise in any setting where the goals of a principal and his/her agent conflict

9. Agency costs for innovative firms PrincipalAgent Agency cost ownermanager risk aversion; preference for “easy life” minority shareholder majority shareholder private benefits preferred to share value maximization VC firm entrepreneur diversion of funds; overconfidence

10. Testing for financing constraints due to info asymmetry (1) Information asymmetry implies internal funds are cheaper than external funds Test: set up R&D investment equation and test for “excess” sensitivity to cash flow shock, as is done for investment Models used: Accelerator - optimal capital stock proportional to output, actual capital takes time to adjust Accelerator - optimal capital stock proportional to output, actual capital takes time to adjust Euler equation - from dynamic program of a value-maximizing firm Euler equation - from dynamic program of a value-maximizing firm

13. Testing for financial constraints due to info asymmetry (2) Three empirical methods: 1. Stratify firms by cash constraints faced, estimate equation for each group and test for differences Stratification sometimes debatable (Fazzari et al vs Kaplan-Zingales) 2. Include a cash flow measure in the equation to proxy for constraints and test for its presence Identification of demand vs supply shocks is an issue 3. Euler equation: model shadow value of investment funds as a function of changes in financial position to test whether it varies with them Theory relies on substitution of financing and spending in adjacent periods, low power

14. Results of tests Methods applied to large firms in the US, UK, France, Germany, Ireland, and Japan Cash flow sensitivity is greater in Anglo- Saxon economies (US, UK, Ireland), although some of the effect may be a response to demand shocks Cash flow sensitivity is greater in Anglo- Saxon economies (US, UK, Ireland), although some of the effect may be a response to demand shocks Low, but not zero, in France, Germany, Japan Low, but not zero, in France, Germany, Japan Greater for smaller firms Greater for smaller firms

15. Testing for agency costs Previous approach will not work – firm is assumed to maximize something other than value Use marginal condition and add indicators of owner-manager separation? Use marginal condition and add indicators of owner-manager separation? Usual method - measure effects of increasing managerial security or the managerial share of firm Usual method - measure effects of increasing managerial security or the managerial share of firm Examine differences in investment behavior across different ownership classes Examine differences in investment behavior across different ownership classes

16. Do agency costs matter? Managers-owners: tests are somewhat weak and evidence is unclear (and magnitude unknown): Antitakeover amendments do not reduce and may increase R&D (US) Antitakeover amendments do not reduce and may increase R&D (US) Institutional ownership associated with higher R&D (US) Institutional ownership associated with higher R&D (US) Diffusely held firms less innovative (measured by R&D spending) (US & UK) Diffusely held firms less innovative (measured by R&D spending) (US & UK) Majority-minority shareholders (Europe) Hall and Oriani – R&D in majority-controlled firms valued less (essentially zero in Italy) – one manifestation of “tunnelling”? Hall and Oriani – R&D in majority-controlled firms valued less (essentially zero in Italy) – one manifestation of “tunnelling”? Munari, Oriani, and Sobrero – family-controlled firms do less R&D Munari, Oriani, and Sobrero – family-controlled firms do less R&D

17. Policy implications? General: Evidence does not contradict Arrow’s argument that there will be underinvestment without some policy attention Evidence does not contradict Arrow’s argument that there will be underinvestment without some policy attention Small and new firms are more disadvantaged Small and new firms are more disadvantaged Agency costs exist, but the story is incomplete – no obvious policy recommendation Gilson: creating a VC market requires three players (VC, investor, entrepreneurs) and correct incentives among them creating a VC market requires three players (VC, investor, entrepreneurs) and correct incentives among them