Intangible Capital’s Impact on Investment-Q Relation
Intangible Capital and the Investment-Q Relation
Based on the research by Peters and Taylor (2016).
Introduction: Shifting Investment Landscape
The neoclassical theory of investment was developed when firms primarily owned physical assets. Consequently, empirical tests of this theory have historically focused on physical capital. However, the U.S. economy has since shifted toward service and technology-based industries, where intangible assets have become increasingly important.
The Core Research Question
Is the neoclassical theory of investment still relevant in an economy increasingly dominated by intangible capital? Specifically, how effectively does Tobin’s Q (Q) explain investment behavior in this new landscape?
Key Findings and Contributions
The authors discover that the classic Q, despite its original design for physical investment, also effectively explains intangible investment. A significant contribution of this paper is the introduction of a new Q proxy that explicitly accounts for intangible capital. This novel proxy demonstrates superior ability in capturing firms’ investment opportunities compared to other widely used proxies.
Detailed Main Findings
- Even though the classical Q performs at least as well for intangible capital as for physical capital, its explanatory power (measured by R-squared values) typically decreases when intangible capital is excluded.
- The within-firm correlation between physical and intangible investment is 31%, but this correlation drops to 17% after controlling for total Q.
- Regardless of how investment is measured, total Q proves more effective at explaining physical, intangible, and total investment, as well as R&D investment and Capital Expenditures (CAPEX).
Methodology and Bias Correction
Recognizing that Ordinary Least Squares (OLS) can yield biased estimates, the authors employ the cumulant estimator. This method produces unbiased slopes and helps measure the proximity of the Q proxy to the true, unobservable Q. Their analysis reveals that when intangible capital is included in the investment-Q regression, the Tau-squared value significantly increases, indicating that their new Q proxy is closer to the true Q.
Capital Adjustment Costs Analysis
Slope coefficients of investment on total Q are instrumental in measuring capital adjustment costs. Specifically, the inverse Q-slope for physical (or intangible) investment quantifies the convex component of physical (or intangible) capital’s adjustment costs. A key finding is that intangible capital’s convex adjustment costs are twice as large as those for physical capital.
Classic Q Theory Fit and Aggregate Results
The authors observe that the classic Q theory fits data better in firms and years characterized by higher intangible capital. However, the underlying reasons for this phenomenon remain unclear, as there is no evidence suggesting that high-intangible firms operate closer to perfect competition or exhibit constant returns to scale. Notably, their main results are even stronger in the aggregate time-series setting, yielding significantly higher R-squared values.
Predictions of the Neoclassical Theory with Intangible Capital
- Prediction 1: Physical and intangible capital share the same marginal Q, and marginal Q equals average Q.
- Prediction 2: Both physical and intangible investment rates, when scaled by total capital, vary with total Q.
- Prediction 3: Total Q helps explain all three investment measures (physical, intangible, and total). It also demonstrates that OLS slopes can identify the adjustment cost parameters, necessitating the inclusion of firm and year fixed effects.
- Prediction 4: Excluding intangible capital from Q generates a downward bias, leading to upward-biased estimates of the adjustment-cost parameter.
- Prediction 5: Assuming physical and intangible capital have the same linear adjustment-cost parameters and purchase prices, a firm will hold relatively less intangible capital if it is costlier to adjust.
Data and Measurement
The authors define a firm’s market value (V) as:
V = Market Value of Equity (MVE) of outstanding equity + Book Value (BV) of debt – Current assets
The replacement cost of intangible capital is calculated as the sum of externally purchased and internally created intangible capital. While external intangible capital is readily identifiable on the balance sheet, internal intangible capital is more challenging to define. The authors accumulate past intangible investment, as reported on firms’ income statements, and define the stock of internal intangible capital as the sum of knowledge capital and organization capital.
The paper provides a detailed discussion of the variables used, which may require some accounting knowledge for full comprehension. For in-depth understanding, referring directly to the paper is recommended.
It is common practice for researchers to discard observations where Q > 10, considering such values unrealistically large. The total Q measure presented in this paper is more relatable, with only 1% of observations exceeding 10.
Full-Sample Results
In their OLS regressions with fixed effects, the authors primarily focus on R-squared values, as coefficients are likely to be biased due to measurement error. Generally, total Q yields significantly higher R-squared values compared to standard Q, with these differences being statistically significant.
A key implication of their theory is that physical and intangible investment should exhibit strong co-movement within firms, given that both capital types share the same marginal productivity and, consequently, the same marginal Q. This co-movement is expected to decrease when the effects of total Q are removed. The authors confirm this by finding that the correlation between the residuals of these two regressions is lower after controlling for total Q. In summary, total Q explains intangible investment slightly better than physical investment, and it explains total investment even more effectively.
Bias-Corrected Results
The necessity of total Q becomes apparent even when considering bias correction using the cumulant estimator. The authors argue that ignoring intangibles introduces a multiplicative, rather than additive, measurement error. Furthermore, the measurement errors from Q and investment are correlated, which violates a key assumption of cumulant estimators. Their analysis demonstrates that applying the cumulant estimator to standard Q versus total Q yields different results.
Interestingly, Selling, General, and Administrative (SG&A) expenses, rather than R&D, show high sensitivity to cash flow. This suggests that bias-corrected results indicate investment is sensitive to cash flow, a finding not explicitly predicted by their theory. A possible explanation for this discrepancy is that SG&A is measured with errors, potentially biasing the cash flow coefficient upward.
Subsample Comparisons
To examine the relationship between total Q and investment across different time periods and industries, the authors conduct subsample tests. A deeper theoretical dive suggests that violating the assumption of quadratic adjustment costs is unlikely to generate the empirical patterns observed in Table 4 (subsampled by intangible capital). While differences in economies of scale or competition could theoretically explain some patterns, empirical evidence supporting these explanations was not found.
Summary of Subsample Findings:
- As firms hold more intangible capital, the explanatory power of total Q increases.
- Interestingly, standard Q can explain both physical and intangible investment, though its explanatory power is less than that of total Q.
- High-tech industries align well with their theory of total Q.
- In later years, Q appears to explain investment opportunities better than in earlier periods.
Alternatives to Adjustment Cost Explanations
Firms utilizing more intangibles tend to have physical capital exhibiting larger convex adjustment costs, and vice versa. The authors consider alternatives to explain differing adjustment costs:
- Differences in purchase prices can explain why some firms use more intangible capital, but they do not explain why firms have different Q-slopes.
- Differences in economies of scale between the two capital types do not necessarily compel firms to use more of one capital type, nor do they cause their Q-slopes to differ.
- Potential candidates for explanation include differences in depreciation rates and adjustment-cost convexities between the two capital types, though these are empirically challenging to demonstrate.
Macroeconomic Results
An advantage of using macroeconomic data is that it does not rely on assumptions regarding the fraction of SG&A representing an investment.
Key Macro Findings:
- Except for a few subperiods, Q generally explains investment relatively poorly at the macro level.
- The relationship between total Q and physical investment remains weak.
- Total Q and intangible investment are strongly related.
- Total investment and total Q also show a strong relationship.
Growth Options and Bond Q
Growth options tend to affect stocks more than bonds, and they influence intangible investment more than physical investment. Consequently, bond Q could serve as a superior proxy for physical capital’s marginal Q, while traditional Q might better measure intangible capital’s marginal Q. Another contributing factor could be that firms with higher intangible investment tend to hold less debt.
In summary, at the macroeconomic level, incorporating intangibles significantly improves Q’s ability to explain investment. However, bond Q still proves more effective at explaining physical investment.
Robustness Checks
The authors’ findings demonstrate robustness across various conditions:
- Total Q explains physical and intangible investment roughly equally well across different sample periods, industries, and levels of intangible capital holdings.
- Total Q consistently serves as a better proxy for true Q, particularly in firms with the highest intangible capital, and this holds true across industry and year subsamples.
- Changes in the lambda value (representing the percentage of intangible investment) do not alter the results, with 30% lambda appearing appropriate.
- Variations in the definition of intangible capital also do not change the core findings.
Key Takeaway
The traditional measure of Q is significantly biased because it fails to account for intangible capital, despite the U.S. economy’s shift towards an intangible investment-driven landscape. Incorporating intangible capital into the Q measure effectively mitigates this measurement error, underscoring the necessity of utilizing total Q moving forward.