This paper is an empirical investigation of corporate R&D expenditures in the UK. It focuses on the main determinants of R&D investment with particular emphasis on the role of institutional shareholders. We follow a methodological approach which is similar to that of Wahal and McConnell (2000). The econometric specification is given by
R&Dit = aindustry + ayear + b1CFit + b2Qit + b3Debtit + b4Equityit + b5InstOwnerit
+ b6 (InstOwnerit * CFit) + eit
where R&D is R&D expenditure, aindustry and ayear are industry and year dummies respectively, CFit is gross cash flow (aftertax income before extraordinary items plus depreciation plus R&D expenditures), Qit is markettobook value of a firm (a proxy for growth opportunities), Debtit is net longterm debt, Equityit is net new stock issuance, InstOwnerit is percentage of institutional shareholdings and InstOwnerit * CFit is an interaction variable.
Our sample consists of 570 firms and 2788 observations covering 15 different industries and a sevenyear time period 2000  2007. Our variables (except for Qit, InstOwnerit, and InstOwnerit * CFit) are scaled down by total assets. This transformation from levels to ratios makes it possible to compare investment and R&D ratios over time and across firms. In a panel with firms that are growing over time as well as starting at different sizes, such a transformation yields trendstationary series and controls for heterogeneity[1].
The summary statistics are presented in Table 1. We find that our sample is highly skewed towards small and mediumsized companies which is suggested by the mean value of total assets being more than 57 times larger than the median value. Hence, in our further analysis we will use median instead of the mean for comparison.

TA (£m) 
R&D* 
PPE* 
CF* 
TD* 
Equity* 
Q 
InstOwn  
Mean 
2042.504 
0.104 
0.148 
0.116 
0.193 
0.116 
2.695 
0.432 

Median 
35.640 
0.038 
0.200 
0.058 
0.106 
0.001 
1.641 
0.425 

St.Dev. 
11438.1 
0.196 
0.185 
0.738 
0.474 
0.294 
4.485 
0.429 

Institutional Ownership 

Year 
2000 
2001 
2002 
2003 
2004 
2005 
2006 
2007 

Mean 
0.365 
0.372 
0.377 
0.388 
0.389 
0.414 
0.565 
0.521 
* variables first scaled by total assets; TA  total assets, PPE  fixed investment, CF cash flow, TD  total debt, Equity  net equity issuance, Q  markettobook ratio
The mean and median percentage of institutional ownership turns out to be 43% for the sample of 4224 observations. There is also an observable trend pattern where institutional ownership is consistently increasing between 2000 and 2006, though in 2007 there is a slight drop, which can be associated with the global financial crisis. This finding of a growing role of institutional investors is consistent with Davis (2002) who analysed the relationship between institutional ownership and corporate sector performance for OECD countries.
Comparing the median values of internal and external financing (internal financing relating to cash flow and external financing relating to total debt and equity issuance) we find that total debt financing is preferred to internal financing (0.106 versus 0.058), while the new equity issuance is the least used method of funding (only 0.001).
Only a small fraction of the total assets, 3.8% (median), goes to R&D which indicates overall low level of involvement with longterm payoff projects. This fraction is very small when compared to 20% (median) going to physical investment. This is expected given the uncertainty of payoffs of R&D projects.
We run 8 different pooledOLS regressions. We include industry dummies to control for differences between firms that arise because of differences in industries that they operate in and time dummies to control for business cycles. We also classify the companies into two categories according to their size; large firms have assets above the median value of �35.639m, while small firms have assets below that median value. The results are presented in Table 2. In the first column we present the findings of the regression for the sample of all firms which includes a stock variable of debt (Total Debtit), while in the second column we present the regression which involves a flow variable of debt (Net Debtit) for comparison purposes. The R2 statistic for regressions that include debt as a stock and flow variable is 0.36 suggesting that replacing total debt with net debt issuance does not affect the explanatory power of the regressions.
We observe a statistically significant negative effect of institutional ownership on R&D for the sample of all firms and for the sample of small firms (0.001 and 0.0007 respectively) suggesting that institutional investors generally exert restraint on R&D investment. This result agrees with Davis (2002), who finds that institutions appear to accompany lower investment in AngloSaxon economies. On the other hand, Wahal & McConnell (2000) document a positive and statistically significant relationship between industryadjusted R&D expenditure and the fraction of shares owned by institutional investors for US companies. Institutional ownership, however, is insignificant in the case of large companies. It may be that institutions in small firms are necessitated to execute the disciplining role in order to increase the value of their investment. For example, Faccio and Lasfer (2000) show that the monitoring role of UK pension funds is concentrated among lowperforming firms and in the long run these firms markedly improve their stock returns. Reasons why such necessity is nonexistent in the case of large firms might be institutions' preference to invest in large firms with more effective management or due to highly dispersed ownership in large firms which creates incentives to take up a passive role instead of an active role. This will be looked at in Section 2.
In the panel of all firms we find that cash flow has a significant negative effect on R&D (0.052) in the regression with the total debt variable. However, this result is not robust when using the net debt as a proxy for debt financing.
Table 2. OLS regressions for R&D investment
Explan. Variables 
All firms (TD) 
All Firms (ND) 
Large firms 
Small firms 

All 
CF < 0 
CF > 0 
All 
CF < 0 
CF > 0 

CFit 
0.052* (3.59) 
0.037 (1.75) 
0.021 (0.7) 
0.049 (1.2) 
0.129 (1.89) 
0.047* (2.75) 
0.021 (0.93) 
0.097 (1.00) 
Qit 
0.009* (4.23) 
0.009* (4.28) 
0.009* (3.66) 
0.033** (2.7) 
0.006* (3.81) 
0.008* (3.91) 
0.008* (3.6) 
0.010* (4.29) 
Total Debtit (TD) 
0.059 (1.49) 
 
0.068* (5.61) 
0.119* (3.92) 
0.030* (2.81) 
0.050 (1.11) 
0.025 (0.45) 
0.056 (1.79) 
Net Debtit (ND) 
 
0.100 (1.39) 
 
 
 
 
 
 
Equityit 
0.028 (1.55) 
0.038* (2.11) 
0.0006 (0.04) 
0.045 (1.56) 
0.020 (1.16) 
0.023 (1.11) 
0.005 (0.21) 
0.036 (1.71) 
InstOwnerit 
0.001* (3.89) 
0.001* (3.73) 
0.0002 (1.75) 
0.0004 (1.02) 
0.0001 (0.98) 
0.0007* (2.57) 
0.001* (2.68) 
0.0005 (1.71) 
InstOwnerit*CFit 
0.001* (3.81) 
0.002* (3.69) 
0.002* (3.05) 
0.003* (2.16) 
0.0006 (0.57) 
0.002* (3.07) 
0.003* (2.82) 
0.002 (0.79) 
No. of obs 
2413 
2400 
1200 
190 
1010 
1213 
699 
514 
R2 
0.36 
0.36 
0.33 
0.45 
0.36 
0.30 
0.26 
0.27 
F statistic 
50.05 
51.38 
23.66 
5.85 
24.99 
21.63 
9.98 
8.17 
Estimated with year and firm dummies (not reported). Heteroskedasticity consistent tstatistics reported in brackets. *Variables are significant at 5% level (tstatistic critical value of 1.96)
Cash flow is statistically significant at 5% level in the sample of all small firms but it is incorrectly signed as we would expect a positive impact on R&D investment. However, the variable loses its statistical significance when we look at small firms with positive and negative cash flows. In the case of large firms, the variable has no impact on R&D. There are some empirical studies which find a significant influence of internal finance on R&D investment; for example, Himmelberg & Petersen (1994) find a large and statistically significant positive impact for small US hightech firms, while Ozkan (2002) confirms the positive relationship for US manufacturing sector. Additionally, Carpenter & Petersen (2002) argue that the typical firm in their sample of publiclytraded US hightech companies finances most of its growth with internal finance. However, our results are consistent with Bond et al (1999) who find that cash flow is not important for the flow of R&D spending for UK companies. Their argument is that cash flow can only predict whether the firm takes up R&D investment not the level of R&D expenditure.
We observe that the interaction variable between cash flow and institutional ownership is statistically significant at 5% level and is negative in all regressions (the coefficient varies from 0.001 to 0.003) except for the regressions for large and small firms with positive cash flows. This negative relationship suggests that the impact of cash flow as a source of financing on R&D is weaker for the firms with higher institutional ownership.
Total debt is insignificant in the panel of all firms and in the panel of small firms confirming the fact that our data set is skewed. Debt is, however, statistically significant and negative for large firms with positive and negative cash flows (0.119 and 0.030 respectively) suggesting that larger firms take into account their leverage when considering an R&D investment decision. The result of the coefficient being negative is consistent with Hall (2002) and Ozkan (2002) who argue that capital structure of R&D intensive firms customarily exhibits considerably less leverage than those in other firms.
Equity is insignificant in all regressions except for the panel with the net debt variable suggesting it is not a major choice of financing. We suspect this might be because equity is costly, especially for small firms. This result agrees with Carpenter & Petersen (2002) who analyse the capital structure of US hightech companies before and after initial public offerings and finds that most firms do not continue to make heavy use of external equity finance after they go public, they tend to rely on internal financing instead.
Tobin's Q is positive and statistically significant in all panels (the tstatistics are greater than 3.60 in most of our regressions suggesting Q is significant even at 1% level). High market's expectations about future profitability of an R&D intense firm tend to boost R&D expenditure. The finding is consistent with Ozkan (2002).
In summary, the pooledOLS results show that significant determinants of R&D investment are institutional ownership, Q and total debt for large firms. Institutional ownership has a significantly negative effect, Q has a significantly positive effect and total debt has a significantly negative effect for large firms. Cash flow is significant at 5% level and it is incorrectly signed for small firms, which causes us to question this result. Net equity issuance has also no effect on R&D investment. Fstatistics in all regressions are very high suggesting that coefficients are highly jointly significant and R2 is also relatively high in all regressions if compared to other similar empirical analyses.
In this section, I am going to discuss the importance of Research & Development (R&D) investment, recent trends seen within R&D investment globally, and the role of institutional investors in determining the level of R&D in firms, do they exacerbate corporate myopia as some suggest or do they encourage long term investment within firms. Characteristics of R&D investment may lead to financing constraints due to capital market imperfections such as asymmetric information adverse selection and moral hazard. This section is a literature review about previous empirical work on R&D investment outlining the theoretical motivation of this paper and the empirical predictions.
Economic theory (Solow 1957) points to technical change as the major source of productivity growth in the long run. It is therefore important to understand the behaviour of R&D investment for different firms with differing financial constraints as R&D activities drive innovation and economic growth. R&D investment supplies much of the new knowledge required for economic development, they are knowledge spillover effects from the inventor to other firms as an idea can be imitated. The fact that R&D is non rival in its nature means that there is scope for the economy to reap high returns from a successful R&D project.
The European Commission Industrial R&D Investment Scoreboard released on 16th November 2009 presents information on the top 1000 EU and top 1000 nonEU companies investing the largest sums in research and development. Global corporate investment in R&D grew more slowly in 2008 (6.9%) than in 2007(9.0%), but remains relatively robust despite the financial crisis that began in September 2008. Companies in the EU saw their R&D investment grow by 8.1% outpacing the 5.7% growth rate for companies in the US. According to the Scoreboard, in the world's top ten R&D spenders the EU had two companies Volkswagen and Nokia; the US held five places including Microsoft and General Motors whereas the top spot went to Japanese carmaker Toyota. The pharmaceuticals and biotechnology sector continued to dominate R&D investment growth in both the EU and in the US accounting for 18.9% of the overall R&D investments by all the Scoreboard companies. Please see Figure 1 in appendix.
Some argue that institutional shareholders are myopic and reduce incentives for managers to invest in long term projects such as R&D as they are a dominant force in setting stock prices so they focus on reported short term earnings. "In today's market, there is an increasing shorttermism driven by institutional investors. You are never better than your last quarter," says B�rje Ekholm, chief executive of Investor, the investment vehicle of Sweden's Wallenberg family. However some argue that institutions may have an information advantage relative to individual investors so are less likely to judge corporate managers on short term reported earnings. Wahal and McConnell (2000) argue institutional investors therefore act as a "buffer" between impatient individual investors and corporate managers allowing managers to focus on projects with long term payoffs. They find a positive relation between R&D expenditures and the fraction of shares held by institutional shareholders. Majumdar and Nagarajan (1997) also find that high institutional investor ownership does not lead to short term behaviour and in particular does not lead to cuts in R&D spending. This follows in the same line of reasoning as Brian Bushee (1998) who finds that corporate managers are less likely to cut R&D expenditures to reverse an earnings decline when institutional ownership is high implying that institutional shareholders are sophisticated investors who have a monitoring role in reducing pressure for myopic behaviour in firms. Francis and Smith (2005) argue "diffusely held firms are less innovative" this implies that monitoring that comes with institutional investors alleviate agency costs and enables investment in innovation (see Hall 2002).
It is important to understand the unique features and characteristics of R&D that make it different to plant, property and equipment investment in order to understand its financing implications. One important feature of R&D investment is that it is a risky investment, there is the extreme uncertainty in outcomes as it difficult to forecast the success of an R&D project at the initiation of the project and its success is by no means guaranteed. Success of an R&D project may be down to timing as your competitors may beat you to the market and an element of luck as your product may not meet the needs of the market. Mansfield et al (1977) report a probability of financial success for R&D projects of only 27%. There is therefore an uncertainty that lies within the future profitability of R&D projects.
Ozkan (2002) points out that the uncertainty involved in R&D projects leads to the fact that R&D investments differ from other types of investments such as physical investment in an important way i.e. R&D investment is not in general collateralizable. R&D projects hold most of their value in the growth opportunities they might offer and the scientific knowledge they possess so they are an intangible asset and have no collateralizable net worth. Hall (2002) points out that banks and other debt holders prefer to use physical assets to secure loans and are reluctant to lend when the project involves substantial R&D investment rather than investment in plant and equipment as uncertainty is high for R&D investments. R&D investment has little salvage in the event of failure whereas if a firm borrows to purchase plant and equipment the lender can be compensated by the value of the physical investment. Carpenter and Petersen (2002) point out that external finance may be expensive because information asymmetries are magnified when assets have low collateral value.
Another characteristic of R&D investment is the high adjustment costs attached to any R&D project due to the high replacement costs of firm specific skilled labour. Bond, Harhoff and Reenen (1999) find that around sixty percent of R&D spending goes on the wages of highly educated scientists and engineers. Their efforts create an intangible asset from which the profitability of the firm will be derived in years to come which is therefore lost if they leave or are fired. Grabowski(1968) points out that research workers cannot be fired and rehired with temporary changes in business conditions because training new workers is expensive and these highly skilled workers possess a great deal of firm specific knowledge that could be transferred to competitors.
Another important feature of R&D investment is that it involves a high level of secrecy between insiders and outsiders of the firm. Firms cannot reveal information about their R&D projects that may be crucial for their success in becoming a market leader or outperforming their competitors. There is therefore substantial information asymmetries between firms and potential investors who may provide financial support for R&D projects because the inventor has better information about the likelihood of success. Himmelberg and Petersen (1994) point out that strategic considerations may induce firms to actively maintain these information asymmetries. Levin et al (1987) report that firms in most industries view patents as an ineffective method of appropriating the returns to R&D and often prefer secrecy. Bhattacharya and Ritter, (1983); Anton and Yao, (1998) observes that this reduces the quality of the signal they make about a potential project which may cause the costs of external finance to increase. The fact that investors cannot make accurate appraisals of value of the R&D project and cannot forecast the probability of success of their investment may cause underinvestment in R&D.
According to the Miller Modigliani theorem (Modigliani and Miller, 1958) optimal levels of investment should be indifferent to a firm's capital structure. Capital structure should be irrelevant to investment because external funds provide a perfect substitute for internal capital under this perfect capital markets assumption. However in the recent literature there is stress on information problems in capital markets and the asymmetry of information between the firm's managers and outside suppliers of finance. Characteristics outlined above will have major implications on the type of financing used for R&D and the provision of financing.
Himmelberg and Petersen (2002) suggest that the marketplace for the financing of R&D is similar to the "market for lemons" model outlined by Arkelof (1970). Arkelof (1970) used this model when describing the market for second hand cars as an example of quality uncertainty. There are good cars "cherries" and defective used cars "lemons". Adverse selection problems arise because the seller has more information than the buyer and the range of actual but unobservable project quality between good firms and "lemons" can be large. This implies that inventors will face a higher cost of external than internal finance for R&D due to the lemons premium. Leland and Pyle (1997) suggest that the lemons premium for R&D will be higher than that for ordinary investment as investors have more difficulty distinguishing good projects from bad when the projects are long term R&D investments than short term or low risk projects. Since 1974 in the US and 1989 in the UK firms had to make disclosures about the level of R&D expenditure, Hall (2002) suggests that the lemons problem can be mitigated due to this transparency but certainly not eliminated.
Most of the literature agrees that R&D firms use little debt in financing of R&D projects see Brown and Petersen (2009), Ozkan (2002), Hall (2002) and Carpenter and Petersen (2002). This is consistent with my findings in Section 1 where debt was insignificant in the panel for all firms. Brown and Petersen(2009) and Carpenter and Petersen(2002) suggest that R&D intensive firms possess this quality due to poor collateral value, adverse selection problems, moral hazard and financial distress. Carpenter and Petersen (2002) suggest that these factors cause the marginal cost of debt to increase rapidly with leverage, meaning when debt is the only available form of external finance, a pronounced funding gap will result as firms will be investing substantially less than they would if debt were a "perfect substitute for internal finance." Carpenter and Petersen (2002) also outlines that extensive use of debt may provide negative expected returns to lenders. They also suggest that debt financing isn't suited to high tech investments such as R&D projects as creditors do not share in firms returns in "good states of nature" unlike equity investors debt holders returns are bounded. Adverse selection problems in debt markets are likely to be most pronounced for high tech investment. This is because of asymmetric information about risk characteristics and default probabilities (see Himmelberg and Petersen, 1994). Debt financing can also lead to moral hazard as compared to lower R&D intensive firms; high R&D intensive firms are likely to have substantial scope for substituting risky projects with low risk projects. Restrictions of risk placed on firms will become more severe as leverage increase due to fears of failure of the project and possibly default. Moral hazard problems thus increase with the degree of leverage, Carpenter and Petersen (2002). Stiglitz and Weiss (1981) note that as interest rates rise, borrowers not monitored by their lenders have an incentive to invest in risky projects which may not be in the best interest of the lender. Borrowers invest in riskier, higher return projects that increase the probability of bankruptcy but may offer no offsetting gain to debt holders if success is achieved. This problem is accentuated as firms become more leveraged. Finally financial distress is likely to increase with leverage; Opler and Titman (1994) find that R&D firms that were leveraged suffered more when facing economic distress because leverage meant that they were unable to sustain R&D programs in the face of reduced cash flow. de Meza and Webb (1987, 1990) present models of asymmetric information in debt markets where the equilibrium outcome is over lending as opposed to the under lending that I emphasise in this paper among with most of the literature. However Carpenter and Petersen (2002) point out that their models do not capture all of the reasons why the cost of debt may increase rapidly with leverage including financial distress.
Most theorectical models of financing constraints assume that equity financing is expensive(see Greenwald and Stiglitz 1993) or unavailable but Blass and Yosha (2001) report that R&D intensive firms listed on the United States stock exchange use highly equity based sources of financing. Hall (2002) suggests that the decision of whether or not to use equity finance is sensitive to the expected rate of return to the R&D project being undertaken as investors require higher returns to compensate for the risk of a "lemon". Carpenter and Petersen(2002) also suggest that equity has a few advantages over debt as a source of financing for R&D investment. Equity investor returns are not bounded unlike bond investors, given the highly skewed returns to R&D investment some firms with successful R&D projetcs will have expected marginal returns that exceed the cost of new quity issues. Furthermore equity does not require the need for collateral and additional equity financing does not increase the probability of financial distress. Hall (2002) points out that in the most extreme versions of the market for lemons model the market for R&D projects may disappear entirely and venture capital are viewed as a solution to this missing market problem. There could be capital market imperfections for equity however, Lee et al. (1996) estimates large issue costs for Initial public offerings (IPOs). Jenkinson (1990) provides a comparison of the regulations governing the IPO process in the United States and United Kingdom and he reports smaller direct costs in the United Kingdom.
The assumption that R&D investments must be primarily funded by cash flow due to the existence of information asymmetries between firms and suppliers of external finance, moral hazard and adverse selection problems associated with debt financing and the high required rate of return attached to equity financing would make this a plausible assumption. Cash flow may affect investment spending because of a "financing hierarchy" in which internal funds have a cost advantage over new debt or equity finance (see Carpenter and Petersen (2002) Fig 1 and Fig 2). Cash flow has obvious cost advantages over debt and equity including lower transaction costs and tax advantages. Furthermore asymmetric information makes it very costly for providers of external finance to evaluate the quality of a firm's investment opportunities.
Himmelberg and Petersen (1994) in their panel of 179 small firms in US high tech industries found an economically large and statistically significant relationship between R&D investment and cash flow. They argue that because of capital market imperfections the flow of cash flow is the principal determinant of how small firms acquire technology through R&D. Ozkan (2002) points out however due to the high adjustment costs involved in R&D investment such as the replacement of highly skilled workers R&D can be less sensitive to cash flows.
In this section I will focus on the estimation method and empirical analysis. I will use fixed effects estimation techniques for my empirical analysis. The purpose of my analysis is to determine whether institutional shareholders have any effect on R&D investment. I also try to understand the effects capital market imperfections have on R&D and physical investment by assessing financial constraints.
The empirical model that will be used in this paper is where is R&D investment for firm i in period t, is the fixed effect, represents the year dummies which accounts for business cycles, TA is the book value of total assets, CF is cash flow which defined as gross cash flow which is net cash flow after tax income before extraordinary items plus depreciation and amortisation plus R&D expenditure. NetDebt measures net debt issuance, Equity measures net new funds from stock issues, Q is an empirical proxy for Tobin's Q and InstOwn measures the percentage of share holdings by institutional investors and InstOwn*CF is an interaction variable between cash flow and the share of institutional shareholders. For the Property Plant and Equipment (PPE) regressions I use the same explanatory variables just replacing the dependent variable with.
The sample summary statistics are based on annual firm observations, I focus my discussion on differences in R&D and determinants of R&D between financially constrained and financially unconstrained firms and also differences in R&D in different industries between the years 20002007. As discussed previously in Section 1 all variables are scaled down by total assets except for Q and institutional ownership.
Table 3A reports some descriptive statistics for R&D, physical investment and the share of institutional ownership. For the full sample of firms the mean and median R&D ratio is steadily increasing from 20002003 and gradually starts declining for the remaining years picking up in 2005  2006 and dropping again in 2007 this can be attributable to the financial crisis. However this decline is also noticed with physical investment as physical investment gradually declined from 20002007. The magnitude of investment is significant larger for physical investment compared to R&D investment. For example in the year 2000 firms spent on average �6700 on R&D investment but �25,300 on physical investment. Share of institutional ownership on the other hand continues to increase through the sample with the mean share of institutional ownership increasing by 43% for the 8 years this could suggest that institutional ownership isn't related to R&D expenditures as there isn't a matched increase in R&D expenditures. We found a negatively significant coefficient for the share of institutional shareholders so the summary statistics are consistent with the OLS pooled regressions.
TABLE 3A DESCRIPTIVE STATISTICS
2000 
2001 
2002 
2003 
2004 
2005 
2006 
2007 

R&D/TA £millions 
Mean Median 
0.067 0.023 
0.080 0.032 
0.094 0.041 
0.130 0.049 
0.123 0.047 
0.105 0.042 
0.108 0.034 
0.099 0.032 

PPE/TA £millions 
Mean Median 
0.253 0.219 
0.229 0.182 
0.217 0.175 
0.214 0.169 
0.204 0.148 
0.181 0.125 
0.163 0.110 
0.173 0.101 

Inst Own (%) 
Mean Median 
36.5 36.37 
37.2 36.51 
37.7 39.98 
38.8 39.03 
38.9 41.32 
41.4 43.01 
56.5 54.95 
52.1 51.60 
Table 3B reports summary statistics for financially constrained and financially unconstrained firms classified by their total debt scaled down by total assets. I have classified firms as financially constrained if their total debt ratio is larger than the median and financially unconstrained if their total debt ratio is lower than the median. A financially constrained firm views external sources of finance as more expensive than cash flow. I use total debt to classify the firms because debt is proxy for leverage and provides a good signal of creditworthiness to investors. Existing debt commitments can influence a firm's access to credit if their debt commitments are high which can therefore limit their R&D investment. A firm with a high total debt ratio is more likely to be financially constrained. However a firm with low debt commitments will have less difficulty in meeting its obligations and would have room to invest in long term projects such as R&D and is financially unconstrained. One can expect that financially constrained firms with high debt ratios will exhibit greater sensitivity to cash flow as they are less likely to find external sources of financing due to their high leverage.
TABLE 3B SUMMARY STATISTICS

Statistic 
R&Dit* 
PPEit* 
CFit* 
TDit* 
NDit* 
Equity* 
Q 
InstOwn 
F. Cons 
Mean 
0.073 
0.265 
0.083 
0.363 
0.021 
0.069 
2.137 
47.06 
Median 
0.021 
0.221 
0.072 
0.260 
0.000 
0.001 
1.511 
46.28 

F.Uncons 
Mean 
0.135 
0.135 
0.149 
0.024 
0.015 
0.162 
3.237 
39.75 
Median 
0.069 
0.083 
0.020 
0.004 
0.000 
0.003 
1.954 
38.54 
Note: * variables first scaled by total assets
F.Cons  Financially Constrained , FUncons  Financially Unconstrained
Table 3B shows that financially constrained firms have lower mean and median R&D ratios than financially unconstrained firms as expected. The mean R&D for financially constrained firms is 0.073 compared to 0.135 for financially unconstrained firms. The median cash flow is higher for financially constrained firms as any investment is more sensitive to internal sources of financing rather than external sources due to higher leverage. Q which is a proxy for growth opportunities is higher for financially unconstrained firms as they have the access to external capital that can allow them to grow and take on additional investment. Equity is higher for financially unconstrained firms as expected as they have access to external sources of financing due to their low leverage. Total debt and net debt are lower for financially unconstrained firms which may suggest that debt is not a financing instrument for R&D as the increase in R&D for financially unconstrained firms is not matched with an increase in debt. The median value for net debt is 0 for financially constrained and unconstrained firms suggesting that the sample has firms with negative values of debt. The PPE ratio is higher for financially constrained firms suggesting that financial constraints may not have an effect on physical investment.
Figure 1 shows the differing levels of average R&D expenditures within different a selection of industries through the years 20002007. As discussed in Section 2 the Pharmaceuticals sector dominated R&D growth globally this year for the EU and non EU Scoreboard companies, this is also evident in this data set as the Pharmaceutical sector dominates R&D expenditures in comparison to the other industries. The Retail industry has the least amount of R&D expenditure of all the industries followed by the Automotive industry then the Computer and Electrical industry. R&D in the Automotive industry severely declined between 20062007 this can be attributable to the financial crisis which severely affected that sector.
There are unobservable firm specific effects in our sample that may influence the level of R&D expenditure within a specific firm i.e. managerial style or the culture of the firm. Controlling for unobservable firm effects is important since the firm effect is likely to be positively correlated with some of the explanatory variables in the model including cash flow and Q. Firms that differ with respect to managerial style means that some managers may want to generate higher cash flows and seek to expand their firms faster than others. See Mundlak (1978) and Hsiao (1986).These firm specific effects are constant over time as managerial style or culture of the company is unlikely to change over the years of our sample (20002007). The fixed effects model is useful in controlling for unobserved heterogeneity when this heterogeneity is constant over time and correlated with independent variables as it allows for these time constant effects to be removed. Failing to account for these firm effects can be viewed as a specification error that is likely to bias the estimate of the effect of our variables on R&D, the effect of cash flow on R&D for example.
The OLS pooled regression used in section 1 failed to remove this heterogeneity so fixed effects as an estimation techniques is more consistent for our sample(see Figure 3 in appendix for comparison between OLS and FE results). The model given in equation (1) where allows for the possibility of firm specific effects, the disturbance term accommodates measurement error in the dependent variable and the effect of unobservable explanatory variables are assumed to be uncorrelated with the observable explanatory variables. We can sum the observations of each firm over the time dimension and divide by T. where Stata uses the standard method of sweeping out the fixed effects by transforming variables to deviations from their firmspecific means i.e. using the within estimator. These firm specific means are calculated by Stata from our sample by taking means for each variable for each firm over the 7 year period.
Equation 3 shows that the fixed effect model eliminates the firm specific error component as the have been dropped by taking the deviations of individual specific time means from each observation. It is called the "within" estimator because it relies on
variations within individuals rather than between individuals. Because fixed effects estimators depend on deviations from their group means, they are sometimes referred to as within group estimators (see Davidson and MacKinnon, 1993). The fixed effects estimator allows for the fixed effect to be correlated with the explanatory variables in any time period but the error term should be uncorrelated with each explanatory variable across all time periods for the fixed effects estimator to be unbiased. There is a loss of degrees of freedom in the fixed effect model due to the time demeaning but Stata corrects for this automatically using the xtreg,fe command.
Estimating equation (2) by OLS gives the between estimator, it uses the time averages for both R&D and the explanatory variables and runs a cross sectional regression. The between estimator is less efficient than the within estimator as it only has N observations instead of N*T observations like the within estimator (see David Winter Lecture Notes 2009 pg14). The between estimator also ignores important information on how the variables change over time. However the between estimator might be of interest as any substantial difference between the within estimator and the between estimator can be regarded as evidence of misspecification in the model. Fixed effects is a common estimation technique in the R&D literature see Himmelberg and Petersen (1994), Fazzari and Petersen (1993) and Carter Bloch (2005).
The fixed effects estimation method allows for individual firm effects, but does not control for potential bias with the explanatory variables i.e. there is an assumption of no correlation between the explanatory variables and the error term which is unlikely to be justified in this model which is why I go on to talk about alternative estimation techniques that can be used later on in this paper.
I perform a Hausman test to test for endogeneity i.e. if the cross sectional effects fixed or random are correlated with the explanatory variables. It is a common test to show whether fixed effects is preferred over the random effects estimator.
RE estimator is consistent & efficient, FE estimator is consistent & inefficient.
RE estimator is inconsistent & inefficient, FE estimator is consistent & efficient
Under the null hypotheses H0 there is endogeneity and the cross section effect is correlated with the explanatory variables. Under the alternative hypothesis H1 the cross section effect is uncorrelated with the explanatory variables. I ran a Hausman test for the main R&D regression for all firms i.e. equation (1).
At the 5% level the critical value of 312.92 is greater than the chi squared critical value at 12 degrees of freedom of 21.026 and the p value if less than 0.05 so we reject the null of no endogeneity. Please see Figure 2 in appendix for the full stata report for the Hausman test. The cross sectional effect is correlated with the explanatory variables and fixed effects (FE) is consistent and efficient. This proves that fixed effects is the preferred estimation technique to random effects for this data set. The FE estimator allows for correlation between and the explanatory variables in any time period as any variable that is constant over time in this case managerial style or firm culture gets removed by the fixed effects transformation in equation (3). This is why the FE estimator is consistent in both the null and alternative hypothesis. The RE estimator on the other hand assumes that the unobserved effect is uncorrelated with each explanatory variable hence why it is inconsistent under H1.
Following from section 1 where I used the robust command to correct for heteroskedasticity I didn't show that heteroskedasticity was present. Using the main regression for R&D, equation (1) I ran the Wald test to test for heteroskedasticity using the stata command xttest3. The error process may be homoskedastic within crosssectional units, but its variance may differ across units a condition known as group wise heteroskedasticity. The xttest3 command calculates a modified Wald statistic for group wise heteroskedasticity in the residuals of a fixed effects regression model, (see page 598 of Greene (2000)).
Homoskedasticity variance of the residuals is homogenous
Heteroskedasticity, variance of the residuals is heterogeneous
The p value is 0.0000 which is significantly less than 0.05 and there is a significantly large critical value (chi2 (517) = 1.7e+37) therefore I can reject the null of homoskedasticity; I therefore use the robust command in all my regressions to account for this heteroskedasticity.
In section 1 I used total debt as my proxy variable for debt; I have replaced total debt with net debt in this section because net debt gives a better indicator of debt you issued in the current period, it is a flow variable whereas total debt is an accumulation of debt i.e. a stock variable. Net debt will be more efficient in explaining the relationship between debt finance and investment. I report F statistics which test the joint significance of all the variables in the regression the null is that the variables are jointly insignificant and the alternative hypothesis is the variables are jointly significant. This test alone does not show which variables has an effect on R&D investment, they may all affect R&D or maybe only one affects R&D. If the null is not rejected however then this justifies dropping the variables form the model.
Following on from the summary statistics presented in table 3B I investigate the impact financial constraints has on investment. The summary statistics showed that cash flow was sensitive to financial constraints as the median cash flow was higher for financially constrained firms. In my econometric analysis I test whether the financial condition of a company has an impact on R&D investment and physical investment. I classify firms according to leverage, higher leverage increasing the likelihood of being financially constrained and having less access to external capital and thus increasing leverage should increase the sensitivity of cash flow on R&D expenditures due to unavailability of external finance.
TABLE 4 FINANCIAL CONSTRAINTS
R&D 
PPE 

Explanatory Variables 
All Firms 
Financially Constrained Total Debt / TA > 50thpercentile 
Financially Unconstrained Total Debt / TA < 50thpercentile 
All Firms 
Financially Constrained Total Debt / TA > 50thpercentile 
Financially Unconstrained Total Debt / TA < 50thpercentile 

CFit 
0.035** (1.81) 
0.024 (1.42) 
0.043 (1.48) 
0.0001 (0.03) 
0.0002 (0.05) 
0.002 (0.24) 

Qit 
0.005* (3.52) 
0.003 (0.60) 
0.004* (2.00) 
0.001* (2.51) 
0.001 (0.88) 
0.001* (2.49) 

Net Debtit (ND) 
0.045 (0.77) 
0.024 (0.62) 
0.004 (0.08) 
0.064** (1.67) 
0.025 (1.61) 
0.028 (1.19) 

Equityit 
0.052* (2.94) 
0.022 (0.58) 
0.059* (3.01) 
0.023* (3.10) 
0.012 (0.79) 
0.025* (3.80) 

InstOwnerit 
0.001* (2.91) 
0.0003 (1.47) 
0.001* (2.33) 
0.0002* (2.09) 
0.0002 (1.63) 
0.0002* (1.88) 

InstOwnerit * CFit 
0.001* (2.67) 
0.001 (1.43) 
0.002* (2.19) 
0.0004* (2.66) 
0.0003** (1.93) 
0.0003 (1.20) 

No. of obs 
2400 
1142 
1258 
2398 
1141 
1257 

R2(within) 
0.26 
0.16 
0.33 
0.12 
0.15 
0.08 

F statistic 
9.07*** 
1.86*** 
6.3*** 
11.32*** 
8.84*** 
4.05*** 
Note: Estimated with year dummies (not reported).
Heteroskedasticity consistent t statistics reported in brackets, see White (1980)
* Variables are significant at 5% level. *T statistic Critical value of 1.96;
**Variables are significant at 10% level. ** T statistic Critical value of 1.64
***Significant F statistic Critical value of 1.5
The share of institutional shareholders is negatively significant for all firms and financially constrained firms that invest in R&D and PPE. This suggests that according to this sample of firms institutional shareholders have a negative effect on R&D investment as well as physical investment in the UK causing managers to have a myopic view towards investment. This result is consistent with the findings of Davis (2002), who finds that institutions appear to accompany lower investment in AngloSaxon economies. The share of institutional shareholders for financially constrained firms is statistically insignificant suggesting that institutional ownership doesn't play a role in investment in highly leveraged financially constrained firms. On the other hand Wahal & McConnell (2000) when looking at 2500 US firms between 19881994 report a positive and statistically significant relationship between industry adjusted R&D expenditure and the fraction of shares owned by institutional investors for US companies suggesting that institutions allow managers to invest in projects with long term payoffs such as R&D. I ran some industry adjusted R&D regressions looking at the different industries reported in the summary statistics and even when looking at different industries institutional shareholders is either insignificant or remains negatively significant for R&D investment. For the Chemicals Healthcare and Pharmaceutical industry (industry 4) the coefficient for institutional shareholders was (0.001) with a t statistic of (1.22), the share of institutional shareholders has no effect on R&D investment in this industry although it is the industry with the highest level of R&D expenditures from the four industries presented in the summary statistics. The second largest investor in R&D is the Computer, Electrical and Electronic Industry (industry 5) which reports a coefficient of (0.0005) and a t statistic of (1.67). There is negatively significant and economically insignificant effect of institutional shareholders on R&D investment at the 10% level of significance.
The interaction variable between cash flow and institutional shareholders is negatively significant for all firms and financially unconstrained firms in the R&D regressions. This negative relationship suggests that the impact of cash flow as a source of financing for R&D investment is weaker for firms with higher institutional ownership. The interaction variable is also negatively significant for all firms and financially unconstrained firms in the PPE regressions suggesting that cash flow as a source of financing for PPE is also weaker for firms with a higher share of institutional shareholders. Q is insignificant for firms that are financially constrained this is plausible as firms with limited access to external finance have less growth prospects. Q is however positively significant for all firms and financially unconstrained firms suggesting that high market's expectations about future profitability of an R&D intense firm will increase R&D expenditure see Ozkan(2002).
Cash flow is insignificant for all the regressions except the R&D regression with all firms where cash flow has a negative effect on R&D expenditures. This finding is unexpected as one would expect cash flow to be more sensitive to R&D in financially constrained firms as this is the only available source of financing. Most of the finance literature that suggest cash flow should be the cheapest source of financing and should have a positive effect on R&D. Ozkan(2002), Himmelberg and Petersen(1994) and Brown and Petersen(2009) all report positive and statistically significant coefficients for cash flow. This result is consistent with Kaplan and Zingales (1997) however, who argue that there is no reason for cash flow sensitivities to increase with the degree of financial constraints as cash flow may act as a proxy for future investment opportunities not captured by Tobin's Q.
Net debt is only positively significant for all firms in the PPE regressions suggesting that debt is a source of financing for firms investing in PPE but has no role in the financing of R&D investment. This is consistent with most of the literature that argue debt isn't a suitable instrument for R&D investment [see Brown and Petersen (2009), Ozkan (2002), Hall (1992, 2002), Carpenter and Petersen (2002) and Himmelberg and Petersen (1994)]. Adverse selection and moral hazard problems are compounded by the lack of collateral value for most R&D investments making debt the least used method of financing. Equity is negative and statistically significant in the panel for all firms and in the panel for financially unconstrained in both the R&D and PPE regressions suggesting that equity has a negative effect on both types of investment. Equity is insignificant for financially constrained firms as expected as they have limited access to external sources of finance.
All regressions have an F statistic greater than 4 meaning that all variables in all the regressions are jointly significant. R2 is the lowest for the PPE regression for financially unconstrained firms (0.08) but in all the other regressions maintains a relatively high power of explanation of R&D. There is also no discrepancy however between the statistically significant negative effect of institutional shareholders on R&D for all firms between OLS pooled and fixed effects estimation (see Figure 3 in appendix). R2 is much higher in the OLS regression due to the industry dummies included; the extra explanatory variables increase the power of explanation of the variation in R&D in the OLS regression. The industry dummies drop out in the fixed effects regression due to the deviations of time demeaned data from the explanatory variables.
I present the between firm results because although this estimator fails to control for unobservable firm effects, Himmelberg and Petersen (1994) suggest that the within firm estimates are biased downwards because of the unresponsiveness of R&D to the transitory component of cash flow due to high R&D adjustment costs and the fact that the transitory component of cash flow averages out over time. The between estimator provides evidence to the extent to which the within firm estimator are biased downward. It is also viewed that any large discrepancies between the within estimator and between estimator can be seen as a misspecification of the model. The between firm estimates are obtained by regressing the firm specific means of the dependent variable on the firm specific means of the explanatory variables i.e. applying OLS to equation (2).
TABLE 5 THE BETWEEN ESTIMATOR

PPE 

Explanatory Variables 
Between Estimator 
Within Estimator 
Between Estimator 
Within Estimator 
CFit 
0.021 (1.31) 
0.035* (1.81) 
0.044* (1.94) 
0.0001 (0.03) 
Qit 
0.013* (8.58) 
0.005* (3.52) 
0.006* (2.84) 
0.001* (2.51) 
Net Debtit (ND) 
0.157* (2.25) 
0.045 (0.77) 
0.002 (0.02) 
0.064** (1.67) 
Equityit 
0.108* ((3.93) 
0.052* (2.94) 
0.067* (1.70) 
0.023* (3.10) 
InstOwnerit 
0.0003 (1.47) 
0.001* (2.91) 
0.001* (2.45) 
0.0002* (2.09) 
InstOwnerit * CFit 
0.003* (5.61) 
0.001* (2.67) 
0.002* (3.38) 
0.0004* (2.66) 
No. of obs 
2400 
2400 
2398 
2398 
R2 
0.45 
0.26 
0.14 
0.12 
F statistic 
31.43 
9.07 
6.53 
11.32 
Note: Estimated with year dummies (not reported).
Heteroskedasticity consistent t statistics reported in brackets for within estimation
* Variables are significant at 5% level; T statistic Critical value of 1.96;
**Variables are significant at 10% level. ** T statistic Critical value of 1.64
*** Significant F statistic Critical value of 1.5
The between estimates show that equity and debt are positively significant for R&D investment whereas with the within firm estimates show that debt is insignificant and equity is negatively significant. Institutional ownership is insignificant in the between firm estimates and cash flow is insignificant. The discrepancies between the within firm estimates is evidence of misspecification in the model. Instrumental variables could be used to reconcile the difference in magnitude between the within firm and between firm estimates (see Himmelberg and Petersen (1994), pg 49). However finding valid instruments for the endogenous regressors may prove difficult using conventional instrumental variables techniques due to possible correlation between the instruments and R&D expenditures, suggesting the use of more advanced alternative methods of estimation. R2 is much higher in the between estimation as the firm specific effect has not been removed.
The estimation of Equation (1) raises several econometric issues as the assumption of zero correlation between the explanatory variables and the error term necessary for the consistency of the fixed effects estimates is unlikely to be justified. Potential endogeneity may arise with cash flow as shocks to R&D investments may also be assumed to affect cash flow. Endogeneity may also arise with institutional ownership as it could be that a high share of institutional ownership causes firms to spend more or less on R&D or alternatively it could be that institutional investors are attracted to firms with high or low R&D expenditures (see Wahal and McConnell (2001)) This potential correlation between regressors and the error term argues for the use of Generalized Method of Moments (GMM) an estimation method however that is out of the scope of this course. It can be described as a more general instrumental variables approach but with GMM the instruments need to be predetermined see Hamilton(1993) and Hansen(1982). The GMM estimator uses equations in firstdifferences, from which the firmspecific effects are eliminated by the transformation, and for which endogenous variables lagged two or more periods will be valid instruments (Arrelano and Bond (1991)). The appeal of the GMM method for panel data is due to the weakness in the assumptions necessary to carry it out. GMM does not require assumptions of zero covariance across years or homoskedasticity across firms for efficiency. Furthermore, the standard error estimates that emerge from GMM estimation are also robust to the presence of correlation across equations, heteroskedasticity, and nonnormality. This is a popular estimation technique in the R&D literature see Ozkan (2002) and Brown and Petersen (2009).
Hausman and Taylor (1981) propose an estimator called the Hausman Taylor estimator and it is based upon an instrumental variable estimator which uses both the between and within variation of the strictly exogenous variables as instruments. The individual means of the exogenous regressors are used as instruments for the time invariant regressors that are correlated with the individual effects. This estimation technique could be useful for this model as it captures within firm estimation and allows the examination of sector effects (see Carter Bloch (2005) pg 219).
My results show that institutional ownership plays no role in determining the level of R&D investment in UK firms. This is inconsistent with research done on US firms where there is empirical evidence that an increased share of institutional ownership allows managers to invest in projects with long term payoffs such as R&D investment (Wahal and McConnell (2000) and Brian Bushee (1998). The results of this paper however leave undecided the source of financing for R&D investment as cash flow and equity when significant have a negative effect on R&D and physical investment. A more advanced technique such as GMM may have given more interesting and more consistent estimates of the variables. Fruitful extensions to my research may include using cash flow net of R&D expenditures. Bond et al (1999) find this measure to be more informative about the investment behaviour of British companies than cash flow gross of R&D costs. Gross cash flow can be argued to generate a strong positive endogeneity bias as R&D appears on both the left and right hand side of the equation. Another extension may be to lag the R&D variable in the R&D regression and the PPE regression to get an idea of a long run effect; this would be possible using the GMM estimator as it allows for serial correlation which may result due to lagged variables.
Stata command : tabstat rndratio pperatio cashflow q totaldebt netdebt equity institutional_pcnt_os cfinst , stat(mean , variance)
Variables 
Mean 
Variance 
R&Dit* 
0.104 
0.038 
PPEit 
0.200 
0.034 
CFit 
0.116 
0.544 
Qit 
2.695 
20.113 
Total Debtit (TD) 
0.194 
0.225 
Net Debtit (ND) 
0.003 
0.018 
Equityit 
0.116 
0.086 
InstOwnerit 
43.240 
1841.594 
InstOwner * CFit 
2.101 
718.705 
First regression OLS POOLED
Stata command : reg rndratio year2000 year2001 year2002 year2003 year2004 year2005 year2006 year2007 ind1 ind2 ind3 ind4 ind5 ind6 ind7 ind8 ind9 ind10 ind11 ind12 ind13 ind14 ind15 cashflow q totaldebt equity institutional_pcnt_os cfinst, robust
Last Regression : Between Effects for PPE
xtreg pperatio year2000 year2001 year2002 year2003 year2004 year2005 year2006 year2007 cashflow q netdebt equity institutional_pcnt_os cfinst, be
1. Akerlof, G. (1970). `The market for `lemons': quality and uncertainly and the market mechanism',Quarterly Journal of Economics, vol. 84 (August), pp. 488�500.
2. Anton, James J., and Dennis A. Yao. 1998. "The Sale of Intellectual Property: Strategic Disclosure, Property Rights, and Incomplete Contracts," The Wharton School, University of Pennsylvania: working paper.
3. Arellano, M. and Bond, S. (1988) Dynamic panel data estimation using DPD: a guide for users. Institute for Fiscal Studies, Mimeo
4. Bhattacharya, Sudipto, and Jay R. Ritter. 1983. "Innovation and Communication: Signaling with Partial Disclosure," Review of Economic Studies 50, 33146.
5. Bloch, Carter (2005) "R&D investment and internal finance: the cash flow effect" Economics of Innovation and New Technology, vol. 14, issue 3
6. Brian J. Bushee(1998) "The Influence of Institutional Investors on Myopic R&D Investment Behaviour" The Accounting Review, Vol. 73, No. 3 (Jul., 1998), pp. 305333
7. Bond, Stephen, Dietmar Harhoff, and John Van Reenen. (1999). "Investment, R&D, and Financial Constraints in Britain and Germany," London: Institute of Fiscal Studies Working Paper No. 99/5.
8. Carpenter, Robert E., and Bruce C. Petersen, (2002) Capital market imperfections, hightech investment, and new equity financing, Economic Journal 112, 5472.
9. Davidson Russell and MacKinnon James G. Estimation and Inference in Econometrics(1993), New York, Oxford University Press,
10. de Meza, D. and Webb, D. (1987). `Too much investment: a problem of asymmetric information',Quarterly Journal of Economics, vol. 102 (May), pp. 281�92.
11. de Meza, D. and Webb, D. (1990). `Risk, adverse selection and capital market failure', Economic Journal , vol. 100 (March), pp. 206�14.
12. Fazzari, S., Hubbard, R. and Petersen, B. (1988). `Financing constraints and corporate investment', Financial Research, vol. 19, no. 1, pp. 59�74.
13. Francis, Jennifer, and Abbie Smith. (1995). "Agency Costs and Innovation: Some Empirical Evidence" Journal of Accounting and Economics 19(2/3): 383409.
14. Grabowski, Henry G., "The Determinants of Industrial Research and Development: A Study of the Chemical, Drug and Petroleum Industries," Journal of Political Economy 76 (1968), 292306.
15. Greene, W. 2000. Econometric Analysis. Upper Saddle River, NJ: PrenticeHall
16. Greenwald, B. and Stiglitz, J. (1993). `Financial market imperfections and business cycles', Quarterly
17. Hall, B. (1992) Investment and Research and Development and the Firm Level: Does the Source of Financing Matter? Working Paper 4906, NBER.
18. Hall, B. (2002) The Financing of Research and Development. Oxford Review of Economic Policy, 18, 3551.
19. Hausman, J.A. (1978) Specification Tests in Econometrics. Econometrica, 46, 12511271.
20. Hausman, J.A. and Taylor, W.E. (1981) Panel Data and Unobservable Individual Effects. Econometrica, 49, 13771398.
21. Himmelberg, C. and Petersen, B. (1994). `R&D and internal finance: a panel study of small firms in hightech industries', Review of Economics and Statistics, vol. 76 (February).
22. Hsiao, Cheng, (1986) Analysis of Panel Data (Cambridge: Cambridge University Press.
23. Jenkinson, T. (1990). `Initial public offerings in the United Kingdom, the United States, and Japan', Journal of the Japanese and International Economies, vol. 4 (December), pp. 428�99.
24. Kaplan, S.N. and Zingales, L. (1997) Do InvestmentCash Flow Sensitivities Provide Useful Measures of Financing Constraints? The Quarterly Journal of Economics, 112, 169215
25. Lee, I., Lochhead, S., Ritter, J. and Zhao, Q. (1996). `The costs of raising capital' Journal of Financial Research, vol. 19, no. 1, pp. 59�74.,
26. Leland, Haynes E., and David H. Pyle. 1977. "Informational Asymmetries, Financial Structure, and Financial Intermediation," Journal of Finance 32: 37187.
27. Levin, R., Klevorick, A., Nelson, R. and Winter, S. (1987). `Appropriating the returns from industrial research and development' Brookings Papers on Economic Activity, vol. 3, pp. 783�831.
28. Majumdar, Summit K., and Amerada Nagarajan. (1997). "The Impact of Changing Stock Ownership Patterns in the United States: Theoretical Implications and Some Evidence," Revue d'Economie Industrielle 82: 3954.
29. Mansfield, E., Romeo, A., Villani, E., Wagner, S. and Husic, F. (1977). The Production and Application of New Industrial Technology. New York: Norton
30. Modigliani, Franco, and M. Miller (1958), "The Cost of Capital, Corporation Finance and the Theory of Investment," American Economic Review 48, 261297.
31. Mundlak, Yair (1978), "On the Pooling of Time Series and Cross Section Data," Econometrica 46, 6985.
32. Opler, Tim C., and Sheridan Titman. (1994). "Financial Distress and Corporate Performance," Journal of Finance 49(3): 101540.
33. R. Solow (1957), ''Technical change and the aggregate production function'', Review of
34. Stiglitz, J.E. and Weiss, A. (1981) Credit Rationing in Markets with Imperfect Information. American Economic Review, 71, 393410
35. The 2009 European Commission Industrial R&D Investment Scoreboardhttp://iri.jrc.ec.europa.eu/research/docs/2009/JRC54920.pdf
36. Wahal, S. and J. McConnell (2000) Do institutional investors exacerbate managerial myopia?
Source: Essay UK  http://buystrangestuff.com/freeessays/economics/randdinvestmentandinstitutionalshareholders.php
This Economics essay was submitted to us by a student in order to help you with your studies.
This page has approximately words.
If you use part of this page in your own work, you need to provide a citation, as follows:
Essay UK, R&D investment and institutional shareholders. Available from: <http://buystrangestuff.com/freeessays/economics/randdinvestmentandinstitutionalshareholders.php> [210318].
If you are the original author of this content and no longer wish to have it published on our website then please click on the link below to request removal: