«Department of Accounting and Finance Working Paper Series AF2014/15WP05 ...»
A third and rather recent explanation of the observed high profitability of pharmaceutical firms relates to difficulties of measuring profit and invested capital (equity), as investments in intangible R&D assets are not fully reported on firms’ balances. As R&D expense is the major cost component in the pharmaceutical industry, financial statements of pharmaceutical firms are strongly affected by whether R&D outlays are reported as an expense or an asset. Only a small number of papers empirically examine whether accounting bias affects profitability in the pharmaceutical industry. In Bloch’s (1974) study accounting rates of return are up to 6 p.p.
higher than the calculated “true” economic rates of return. Subsequent research by Ayanian (1975), Clarkson (1977), Grabowski and Mueller (1978), Megna and Mueller (1991) and Mahlich and Yurtoglu (2011) confirm a significant accounting bias on ROE when R&D is expensed rather than capitalized. However, none of the papers elaborated on the determinants of the accounting bias as we are going to do. The sample size was small as well in most of the studies. Clarkson’s (1977) analysis for instance was based on one single company.
Implementation of accounting rules generally leads to expensing a large portion of R&D outlays as incurred, while economically R&D outlays can be seen as an asset. The standard setters pursue this approach as they weigh reliability over relevance in case of R&D investments and argue that future economic benefits from R&D investments cannot be reliably measured.
Thus, conditions for asset recognition are not met by investments in research or development.
For example, U.S. GAAP applicable for U.S. listed pharmaceutical firms bans capitalization of R&D investments (with the exception of software manufacturers); instead R&D outlays are reported as expense.
Expensing of R&D understates the asset base of pharmaceutical firms and their shareholders’ equity (e.g., Salamon, 1982; Salmi, 1982; Taylor, 1999). It also biases profit, as income statement accounts for current R&D outlays instead of economic amortization of previous R&D investments (e.g., Fisher and McGowan, 1983; Salamon, 1985, 1988). Thus, accounting for R&D can bias the nominator (earnings) or denominator (equity) of the ROE. The sign and magnitude of the bias will depend on the magnitude of R&D investments (Rajan et al., 2007). To illustrate this logic, we report in Table 1 a stylized example of two pharmaceutical firms (firm A and B) that invest each year 100 into R&D and generate 115 in annual revenues.
Both firms have a cash asset (30), some investment in R&D, and no liabilities (that is, total assets equal shareholders’ equity). We use different accounting treatments for R&D in case of firm A and firm B to show how R&D accounting affects accounting profitability measures. Firm A capitalizes 100% of R&D outlays and amortizes them over the next year when they contribute to generating revenues. In this case, reported profit and equity are not biased, and accounting return is equal to the economic return. While being economically “correct”, the example of firm A has also an intuitive appeal: The firm advances 100 to generate revenues of 115. Therefore, return on its investment is (115–100)/100 = 11.5%. As firm A has no liabilities, its ROE equals the Return on Assets (ROA) and hence is 11.5%.
Firm B is allowed only to capitalize 10% of its R&D investments and must expense the remaining 90% of R&D investments when incurred. Expensing of larger portion of R&D outlays leads to understatement of assets and equity. Because equity is understated, ROE of firm B is substantially overstated and equals 37.5% in years 1-3 and 262.5% in year 4. Similar, when investments increase, equity is biased downwards and ROE is biased upwards if firm expenses R&D (Lev et al., 2004; Rajan et al., 2007).
The effect of R&D expensing on earnings depends on whether current R&D expenditure exceeds amortization of previously capitalized R&D outlays. To illustrate, we next assume economic shocks and let R&D investments decrease (increase) to 90 (110) in year 3.
Untabulated results reveal that when R&D is expensed as incurred, any decrease in R&D investments during an economic downturn in year 3 has an immediate income-increasing effect (ROE 60%). However, decreasing investments when those investments are capitalized produces an income-increasing effect only in subsequent years when amortization of investments affects earnings. As a result, the gap in reported performance between expensing and capitalizing firms is expected to increase during an economic downturn (60% vs. 11.5%).2 During the periods of economic expansion, an increase in R&D investments depresses earnings of the R&D expensing firm and lowers the magnitude of bias (15% vs. 11.5%).
Unlike firms from other industries, pharmaceutical firms are not allowed to fully
economic shocks: A temporary decrease in investments increases the magnitude of bias, while a temporary increase in investments lowers bias.
Hypothesis 1: Expensing R&D overstates ROE of pharmaceutical firms and the magnitude of this bias increases (decreases) when economy contracts (grows).
The example provided in Table 1 demonstrates how R&D expensing can bias ROE. In the example, this bias leads to an overstatement of accounting returns. On the other hand, one can also identify conditions under which the positive ROE bias reverts, and pharmaceutical firms report ROE that is potentially lower than ROE reported by firms from other industries. We next posit that certain combination of accounting rules and decelerating investments can negatively bias pharmaceutical firms’ ROE.
While U.S. firms follow U.S. GAAP and are required to expense R&D, accounting standards applicable in other countries allow more latitude in capitalizing some R&D outlays (Hung, 2001). Critically, IFRS applicable for listed pharmaceutical firms in Europe and in most other countries requires that research outlays are expensed as incurred, but permits capitalization of development costs. Specifically, IAS 38 requires companies to separate the research phase
In the actual reporting practice, the gap is further going to increase due to reporting of impairment losses by firms that capitalize their investments. Particularly, during an economic downturn firms typically report an impairment of some investments to account for the fact that future cash flows from those investments are expected to decline. In turn, R&D expensing firms carry no (or very limited) R&D assets and thus have no assets to impair, leading to relatively higher income when economy contracts.
from the development phase of a project and to capitalize costs incurred during the development phase if capitalized costs meet a set of criteria such as ability to sell the developed product and ability to measure future economic benefits. In case of the pharmaceutical industry, these criteria are typically met when a drug is approved by the regulatory body. This leads to more development costs capitalized on balances of international firms relative to their U.S.
counterparts. Also, while U.S. GAAP requires R&D purchased in the process of merger or acquisition (in-process R&D) to be expensed, IFRS allows capitalizing in-process R&D. Thus, IFRS financial statements of pharmaceutical firms may be somewhat less affected by accounting bias relative to U.S. GAAP financial statements.
Importantly, capitalization of some R&D investments increases earnings during the periods when companies grow, because larger portion of current R&D outlays are deferred into the future. However, in later years when companies mature and experience declining growth rates in R&D investments capitalization leads to lower profitability of R&D capitalizers (Aboody and Lev, 1998). This is because the amortization of previously capitalized R&D investments is higher than R&D outlays of the current period. Aboody and Lev (1998) support this prediction by showing that declining investment growth rates in software industry led to lower profits of R&D capitalizing firms. The descriptive statistics for our international sample of pharmaceutical firms indicate that our sample firms are mature (average age 36 years) and that some countries experience declining rates of R&D investments. This evidence is consistent with Cincera and Veugelers (2014) who find that U.S. has a greater number of young innovating pharmaceutical firms relative to Europe. We predict that declining growth rates may lead to lower ROE reported
by international pharmaceutical firms:
Hypothesis 2: Decelerating R&D investments of pharmaceutical firms bias ROE downwards.
Anecdotal evidence suggests that the regulation of pharmaceutical industry may be motivated by high profitability of pharmaceutical firms (see e.g., Ecorys, 2009; Pear, 1993; Rapp and Lloyd, 1994). The link between accounting profits and regulation is not unique to the pharmaceutical industry. Abnormally high profits were cited in the congress request to the Federal Trade Commission to investigate possible price-fixing of gasoline by U.S. refineries3 and were argued to cause the intervention of the Federal Trade Commission in the breakfast cereal industry (Scherer, 1982). This anecdotal evidence is supported by a theoretical literature that predicts regulators’ attentiveness to firm performance as an indicator of market power (Watts and Zimmerman, 1986). The economic theory of regulation also recognizes other rationales related to firm performance such as rent distribution, revenue generation and redistribution (Posner, 1971). However, we are not aware of any empirical study that examines whether higher profits lead to regulatory actions. Furthermore, there is no prior evidence on whether in their decision making regulators rationally adjust for any bias resulting from application of accounting rules. If high profitability motivates industry regulation and regulators do not correct for this bias, the resulting regulations might distort firm R&D investments.
Accounting literature shows that users of financial statements are often limited in their understanding of how financial statements are articulated and fixate on the bottom line earnings in their evaluation of financial performance. For example, Sloan (1996) shows that investors focus on aggregate earnings and disregard information in earnings components despite those components being informative about sustainable profitability. Similar, Lev et al. (2004) find that investors do not correct for the bias caused by R&D accounting and as a result firms with abnormally high profitability are overvalued. Jones (1991) shows that firms manipulate earnings
The letter asking for the investigation argued that “[a]t a time when major refiners and oil companies are making record profits and American families continue to struggle with gasoline at record prices, the idea that refiners may be manipulating the market to keep prices artificially high is offensive.” to affect regulatory outcomes and argues that regulators ignore such earnings bias in their regulatory decisions. We predict that regulators are likely to be attentive to high profitability of pharmaceutical firms leading to a greater oversight of the pharmaceutical industry. We also predict that similar to other stakeholders, regulators ignore the bias caused by R&D accounting in their decision making. That is, regulators fixate on aggregate earnings and do not adjust
profitability for any accounting biases. We formulate our predictions as follows:
Hypothesis 3: High profitability of pharmaceutical firms leads to increased regulatory activity.
Hypothesis 4: Regulators fixate on aggregate earnings and do not adjust for the bias caused by R&D accounting.
3. Empirical approach
3.1. Difference in ROE between pharmaceutical and non-pharmaceutical firms and its determinants We conduct our analysis in three steps. First, we benchmark ROE of pharmaceutical firms against ROE of firms from other industries – our proxy for “normal” profits – to examine whether accounting rules for R&D bias profitability of pharmaceutical firms. Second, we adjust ROE of pharmaceutical firms to reveal the magnitude and the sign of accounting bias. Third, we relate accounting profitability of pharmaceutical firms and accounting bias to the regulation of the pharmaceutical industry.
We use two proxies to capture effects of R&D accounting on the ROE difference between pharmaceutical and non-pharmaceutical firms. We use the difference in R&D intensity between pharmaceutical firms and non-pharmaceutical firms to proxy for the effect of R&D accounting on ROE bias. Hypothesis 1 predicts that ROE of pharmaceutical firms will be higher due to expensing of their R&D investments. We use changes in gross domestic product ΔGDP to capture the effect of economic upturns and downturns on investments and reported ROE differences between pharmaceutical firms and firms from other industries. Hypothesis 1 predicts that economic upturns (downturns) will lead to relatively lower (higher) earnings reported by
pharmaceutical firms. This leads us to the following empirical model:
_ &_ ∆ (1) where ROE_diff is the country’s i difference in weighted-average ROE between the sample of pharmaceutical and non-pharmaceutical firms in year t. R&D_diff is the difference in weighted average R&D expenditure (the ratio of R&D expenditure to closing book equity) between the sample of pharmaceutical firms and the sample of non-pharmaceutical firms in a given year.
Weights are set to the share of book equity in the total equity of pharmaceutical firms in each country-year. ΔGDP is the change in GDP over a year in a given country. We predict α1 to be positive and α2 to be negative. We estimate model (1) by means of an OLS regression. We base our inferences on robust standard errors that are adjusted for heteroscedasticity and time-series dependence.4 Additionally, we use country fixed effects to control for country characteristics when we fit model (1) using our international sample.
We test hypothesis 2 using extended version of model (1). We add an indicator variable