«Gareth Macartney University College London, Institute for Fiscal Studies and AIM April 2007 [preliminary- please do not quote without the author’s ...»
Investment in Innovation and Fixed Assets: the effect of US tax
reforms that reduced the relative price of equity to debt
University College London, Institute for Fiscal Studies and AIM
[preliminary- please do not quote without the author’s permission]
Tax reforms in the US in the mid-1980s reduced the relative tax cost of equity and internal
finance to debt by 30%, from a large reduction in the statutory rate. This paper finds evidence
that for innovative firms this lead to a decrease in ordinary investment and an increase in innovative investment, while controlling for other variation in the prices of ordinary and R&D investment. This suggests that equity and internal finance are cheaper for innovation than debt finance, to the extent that exogenous decreases in their relative prices can cause innovation. The ability to substitute from ordinary investment to R&D investment was larger for firms with plentiful internal financial resources, consistent with a large external financing premium for R&D.
JEL codes: L2, G32.
Keywords: Innovation, capital structure, debt, equity, investment.
Acknowledgements: The author would like to thank Sami Berlinski, Nick Bloom, Wendy Carlin, Rachel Griffith and Bronwyn Hall for useful suggestions.
Correspondence: email@example.com; IFS, 7 Ridgmount Street, London WC1E 7AE, UK.
1 Introduction Innovation as investment is long term and risky, involves high information asymmetry between investors and firms due to its specialist nature, and offers little collateral.1 Such attributes imply that innovation requires equity finance (or internal funds if available), rather than debt finance.
Equity instruments involve a residual claim on returns and therefore, whatever the personal risk preferences of investors, such instruments are better suited to risky investments than debt instruments.2 Secondary markets in shares enable equity holders to crystallise gains or loses as more information becomes available. Equity markets involve diverse ownership which promotes hard budget constraints in that re-financing of projects requires attracting new investors.3 Equity holders have some residual control of the firm in the form of voting rights, although it is fair to say that debt holders can exercise control in the form of debt covenants at the time of loan renewal. Such considerations would imply that firms face a lower (unobserved) cost of capital for the R&D project if financed by equity rather than debt.
Empirical evidence to date supports the assertion that internal funds and equity are associated with R&D, see Hall (2002) for a survey. Aghion et al. (2004) shows that for publicly traded firms those that report R&D expenditure are more like to issue new equity and more so with greater R&D intensity. Bond et al. (2003) find that the availability of internal funds predicts R&D participation, although not the R&D intensity thereafter, in UK firms. Although highly informative these results could be driven by omitted variables such as demand shocks that affect both the incentive to innovate and the availability of internal funds or the ability to raise equity.
The current paper implements a new identification strategy to test if, as debt finance becomes more expensive relative to equity and internal funds, firms engage in more R&D. The paper uses changes in the tax treatment of debt relative to equity as an exogenous source of variation in the relative price.
The specific tax reform used is the 1986 corporate tax cut in the US from 49.6 percent to 34.0 percent. As the interest on debt is tax deductible the tax advantage of debt over equity decreases with corporate tax cuts and this tax reform, therefore led to a reduction in the relative price of equity to debt. As part of the same reform the tax base was broadened, leaving the direct effect of taxation on the price of investment largely unchanged, at least on average. This paper controls for variation in the price of investment, and, also controls for the impact of the R&D tax credit of
1981. The results suggest that US manufacturing firms substituted from fixed capital to ‘knowledge’ capital in that the ordinary investment rate decreased and the R&D investment rate increased. Furthermore the substitution to R&D investment is larger for firms with internal funds, as identified by their dividend behaviour, possibly because these firms faced a lower cost of finance for R&D than those firms dependent on external finance.
1 About 50% of R&D costs are spent on the specialised labour of scientists and engineers, Hall (2002).
2The payoff schedule for equity is convex for a firm with debt, whereas for the debt holder it is concave. Equity holders like the possibility of high returns.
3 See Dewatripont and Maskin (1995).
Throughout the paper assumes a hierarchy of finance model of firm investment.4 Under strong assumptions the cost of capital should be independent of the type of capital used: the financial structure of the firm should be irrelevant to the value of the firm.5 These assumptions are: one, there is perfect information between firms and investors; two, there is no difference in the tax treatment of different types of finance and; three, there is no loss in the event of bankruptcy.
Where informational asymmetries do exist all firms must pay investors a ‘lemon’ premium to insure investors against the possibility that they have invested in a bad firm6. This premium increases the cost of external finance above that of internal finance. In all developed economies debt has a tax advantage in that interest payments on debt finance are exempt tax at the corporate tax rate. Bankruptcy risk increases with the indebtedness of the firm as the firm has greater and greater interest payments that it is obliged to meet and, if we assume that there is some intangible loss of firm value in the event of bankruptcy the cost of debt capital increases with indebtedness.
These considerations lead to a hierarchy of finance for ordinary investment where for the marginal investment internal finance is cheaper than external finance, and, due to its tax advantage, debt is cheaper than equity. The aim of this paper is to find empirical evidence that the ordering is different for R&D projects, in that as equity becomes cheaper relative to debt there is greater R&D investment relative to ordinary investment.
As well as clear implications for the nature of the tax treatment of debt this paper has implications for the finance and growth literature. Overall financial development appears to matter for growth in developing countries, but whether this finance is bank based or market based seems to be unimportant in the aggregate, see Levine (2005) for a survey. However, in developed countries Carlin and Mayer (2003) find cross-sectional evidence that inherently equity dependent, skill-intensive industries grow faster, through increased R&D, in countries with good accounting standards, suggesting that informational concerns are the reason behind some industries dependence on certain financial structures.7 The current paper provides direct evidence that institutional structures (in this case tax) that affect firm financial structure are important for an outcome that we have strong evidence to suggest drives productivity growth.
This paper proceeds as follows: section 1 models the effect on ordinary and R&D investment of a reduction in the tax advantage of debt; section 2 describes the empirical strategies and data used; section 3 provides results and a final section concludes.
4 See Bond and Meghir (1994), Bond and Soderbom (2006).
5 See Miller and Modiaglini (1958).
6 As described by Akerloff (1970) for the used car market.
7Or conversely, as is the point of Carlin and Mayer (2003), why some countries with specific financial structures engage in certain activities.
2 Modelling a Shock to the Tax Advantage of Debt For a firm investing in ordinary and knowledge capital, where the former can be financed with a mixture of debt, internal funds and equity, and the latter can be financed only with internal funds and equity, a decrease in the tax advantage of debt will have two effects. Firstly, it will change the steady state mix of factor inputs shifting from ordinary capital to “knowledge” capital. If we assume that the firm has an optimal output level (i.e. the firm has market power), this will lead to lower ordinary capital and higher knowledge capital in the steady state. Secondly, if the firm faces strictly convex adjustment costs a decrease in the tax advantage of debt will affect the optimal investment rate.
Considering the first effect, financing costs are important in the steady state as capital stocks have to be maintained in the face of depreciation and this requires funding. The first effect also assumes that firms have market power so that there is an optimal output level for given variable (labour) costs. If this were not the case any increase in the price of one factor would lead to less output- the factor mix would change only in the sense that ordinary capital would decrease, with knowledge capital likely decreasing also if factors are complementary. So, for some optimal
output level Y ∗ = F ( K ∗, G ∗ ) and linearly homogeneous production we can write:
where Yt, K t and Gt denote output, ordinary capital and knowledge capital respectively. The prices rt K, rtG denote the user costs of capital for K t and Gt respectively. This leads to the
familiar condition for the marginal rate of technical substitution:
and we assume that production is such that the left hand side is more negative for higher Gt relative to K t. From the previous discussion it is assumed that the cost of finance for ordinary investment is increased by a loss in the tax advantage of debt and, so, the steady state value of ordinary capital, K t∗, will decrease, and the steady state value of knowledge capital, Gt∗ will increase. Output will remain at the optimal level, as long as the new factor mix is within the feasible production set (i.e. the change in relative prices is not too large). It is assumed that this will happen in some average sense for all firms and a decrease in the tax advantage of debt is considered an increase in the ratio of the cost of capital for debt finance to the cost of capital for equity and internal finance, expressed as rt Debt rt Equity / Internalfunds and referred to hereafter as the ‘cost ratio’.
The optimal investment rate of a firm facing convex adjustment costs is given by8:
where Π It = ∂Π t ∂I t is strictly positive in the ordinary investment rate, I t K t, and decreases linearly with the price of investment goods. λtK is the shadow value of ordinary capital and α tK is the shadow cost of finance for ordinary capital investments. Assuming convex adjustment
costs for investment in knowledge capital a similar condition holds for R&D investment:
inter-temporal condition dependent on discounted expected future marginal productivity shocks.
The change in the tax advantage of debt will leave α tG unchanged if debt is not used for R&D investment, as we believe.
Assuming that ordinary investment can be finance by a mixture of debt, cash and new equity, but that R&D investment can only be financed by cash or new equity, a reduction in the tax advantage of debt will have the following effects. Firstly, it will change the desired steady state production mix in favour of knowledge capital. Secondly, where it increases the shadow cost of finance, it will reduce the ordinary investment rate. Thirdly, it will increase the R&D investment rate as the firm tries to adjust to the new steady state and to make up for the decrease in the ordinary investment rate, and the effect will be dampened where the shadow cost of financing R&D investment is high. This latter effect comes from both the new steady state and the adjustment of investment, simply in both cases it is more costly to substitute from ordinary investment to R&D investment when the cost of R&D investment is high.
At this stage we can summarise these results as follows:
8 See Hayashi (1982), Bond and Meghir (1994), Bond and Soderbom (2007). Also, Hall (1995) and Hall and Hayashi (1989) for models of dynamic investment with tangibles and intangibles. Investment is productive next period.
Prediction 1: A decrease in the tax advantage of debt will lead to a lower fixed investment rate and a higher R&D investment rate.
Prediction 2: The increase in the R&D investment rate will be lower for firms facing higher financing costs.
We can consider a change in the desired steady state mix of factors from K t∗ to Gt∗ as a reduction in the shadow value of ordinary capital, λtK, and an increase in the shadow value of knowledge capital, λG. Intuitively, ordinary capital becomes less valuable as the financial cost of t maintaining it increases, and as a result, to maintain optimal output, knowledge capital becomes
more valuable. This implies that:
where S t is a measure increasing with the shadow cost of finance, and the CostRatiot increases as debt becomes more expensive relative to other forms of finance, i.e. as its tax advantage decreases. Predictions 1 and 2 suggest that an increase in the cost ratio should be decreasing in (4) and increasing in (5), and that in (5) its increasing should be less for high S t. How changes in the cost ratio will be differential across firms with different measured financing costs warrants further discussion about the hierarchy theory of finance. This and a discussion of what is observable concerning the shadow cost of finance faced by the firm is covered in the next section.