Insights on Productive Investment: Firm-Level Data Analysis

What is productive investment? Insights from Firm level data for the United Kingdom Sudipto Karmakar, Marko Melolinna and Philip Schnattinger Mary O Mahony Kings College London Presentation at the 38th IARIW General Conference Session 7D-1, Productivity/investment

Overview of the Paper Overview of the Paper Studies the effects of types of corporate investment and levels of debt on productivity. Uses firm-level data from the United Kingdom. And presents a stylised model of dynamic firm profit maximisation, augmented with an external financing option. This paper s contribution is to study how leverage and types of investment interact. Main conclusion is that levels of debt are not necessarily bad for TFP, if the debt is accompanied by high levels of productive investment intangibles, mainly R&D.

Related Literature Related Literature Typically, firm-level studies find a negative relationship between high levels of corporate debt and subsequent capital expenditure. Much of this literature is focused on the business cycle, showing poor productivity performance after financial crises of financially vulnerable firms. Similarly, large literature on types of investments and impacts of R&D shocks on productivity growth, Paper by De Ridder (2022) which brings in market power arguments. This paper differs from the previous literature by focusing on the links between types of investment and leverage, rather than the more common business cycle properties of this relationship.

Empirical Analysis: Specification Empirical Analysis: Specification Regress TFP on capital, distinguishing tangible and intangible, and debt, and interactions, including controls for firm characteristics such as firm age, size, cash holdings and profits. Uses OLS panel regression model with lagged explanatory variables to partially account for endogeneity. The interaction term allows them to distinguish between good" and bad" leverage. They hypothesise that leverage per se is not detrimental, if it is used to finance productive investment. To aid interpretation of the interaction term, and focus on high investment firms, the investment variables are 0/1 dummies, with value 1 if the firm's investment ratio (R = tangible or intangible) is in the top quartile of firms for a particular year, and 0 otherwise. They use industry median corrected values for the debt ratio (D) provides a way of controlling for time-varying industry-specific effects results are robust to not including this adjustment Zi,t = 0+ 1Di,t-1 + 2Ri,t-1 + 3Di,t-1*Ri,t-1 + ci + ft + Xi,t-1 + ei,t Where Zit is (log)level of TFP for firm i at time t.

Empirical Analysis: Data Empirical Analysis: Data Worldscope financial account data on UK listed firms. Annual data 1990-2018. Exclude finance and oil sectors (ISIC 2-digit sectors 2, 19, 64 66). 27,712 firm year observations. Data winzorised at 1st/99th quantiles. Intangible capital stocks use a combination of book value intangibles, accumulated R&D and accumulated sales, general and administrative (SGA) spending. Well known issues of calculating TFP from firm data, due to endogeneity of inputs and outputs, so authors use an established method based on a production function approach suggested by Ackerberg et al. (2015), a variant of the Olley- Pakis method. But robust to using the popular Wooldridge (2012) approach.

Empirical Analysis: Summary Statistics Empirical Analysis: Summary Statistics The descriptive statistics show average TFP growth is around 1.3% per annum, in line with estimates of aggregate TFP growth in the UK during the same time. They also find a positive contemporaneous correlation between TFP and the intangibles stocks and flows, and a negative relationship between TFP and tangible investment flows. but the relationship between TFP and debt is more complicated. Overall, these stylised facts confirm the importance of accounting for firm level heterogeneity when looking at the relationships between TFP, investment and debt. Correlations suggest that intangibles stocks (and flows) are higher in firms that are less indebted, younger, smaller, more cash rich and less profitable than those with smaller intangibles stocks. The TFP distribution of high-intangibles firms is higher than that of low intangibles firms, pointing to a between the level of TFP and intangibles. High intangibles firms are heavily concentrated in the manufacturing and ICT sectors.

Empirical Analysis: Results The effects of the intangibles variables on the level of TFP are strongly positive for stocks and flows. In contrast the effects of tangibles are significantly negative

Empirical Analysis: Results Debt on its own is always insignificant Interaction of debt with intangibles stock is significant The total effect of debt with the intangibles variables is significantly positive, as shown by p-value. And not so for tangibles.

Empirical Analysis Empirical Analysis: Economic Size : Economic Size There is a large and significant effect of 9.7% on TFP for those firms that are in the highest quartile of intangibles stock levels. This effect is lower, but significantly positive, for intangibles flows (Consistent with lagged responses) and significantly negative for tangible stocks and flows. The results suggest that a combination of high debt and high "productive" investment can be associated with high levels of TFP.

Empirical Analysis: Main findings Empirical Analysis: Main findings Additional analysis shows that the intangible investment effects are largely driven by R&D and organisational capital. In contrast, balance sheet measures of intangibles are not significant drivers although flows of goodwill capital are weakly significant. Moreover, the positive interactions between debt and high levels of intangibles is purely driven by R&D investment. So it is ultimately the combination of high R&D intensity and high debt that drives the positive debt effects on TFP. Also report several robustness checks.

Structural Model Structural Model Use a structural model to set out the mechanisms between debt, investment in tangible as well as intangible capital. Model details are in the paper The model features a negative effect of higher leverage on the cost of external financing Firms will take the additional cost of external financing into account when choosing the optimal investment expenditure. They further assume that a higher intangible capital share in total capital has the potential to further increase debt cost following evidence presented in Falato et al. (2022) that a higher intangible capital share reduces a firm's capacity to provide collateralizable assets. They then use this model to see if they can replicate the relative relationships between key variables that were found in the empirical section. They solve the value function for the firm's optimal policies given the firm's states and approximate the value function of the firm, given these states, with a neural network. The advantage of this approach is that they are able to estimate parameters, which would be more difficult to calibrate based on empirics.

Structural Model: Estimation Structural Model: Estimation Generally good fit of model (red line) to the conditional data moment (solid blue line).

Structural Model: Estimation Structural Model: Estimation Investment in Physical & Intangible Capital Higher investment in intangible capital (per total assets) is associated with higher productivity, while higher investment in physical capital is not.

Structural Model: Estimation Structural Model: Estimation Investment in Intangible Capital & External Financing Firms that take on more debt and invest to a higher extent in intangible capital have higher productivity.

Structural Model: Estimation Structural Model: Estimation Investment in Tangible Capital & External Financing Firms that take on higher debt and invest more in physical capital will have lower productivity,

Conclusions Conclusions Back of the envelope calculations from the model suggest that lower investment in intangibles can account for about 10% of the TFP slowdown in the UK. Could be more if the investment behaviour of these large firms carried over to other firms. This paper contributes to the long-standing discussions on the effects of innovation, or intangible investment, as well as of debt, on firm performance. Says little about effects across the business cycle (future work). The paper emphasises the importance of understanding what debt is used for, when analysing its effects, an important consideration when setting policies that affect or operate through firms' debt.

Comments Comments Very nice paper combining empirical analysis with a theoretical model, giving plausible results. Use of the term productive investment maybe use productivity enhancing. Defining investment as 0-1 dummies why not use continuous variables? Only partial measures of intangibles in firm level accounts. SG&A spending used as a proxy for organisational capital. Only large listed firms in sample are the results likely to carry over to smaller firms?

Comments Comments given a firm is in the high debt bucket, being in the top decile in terms of intangibles stocks/flows has a strong positive debt effect on TFP . What about negative impacts from market power (De Ridder)?