Market Dominance & Socioeconomic
Weight Dynamics in the US
Pharmaceutical Industry
An extensive view on the
OTC-Analgesics market
hhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhh hhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhh hhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhh hhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhBachelor of Science
Edited byMilan van Aagten
Supervised byStephan Jagau
Abstract
Statement of Authenticity
During the last decades the US pharmaceutical industry has been subject to numerous incidents, challenging its highly self-valued image on trustworthiness and benevolence. This paper investigates how and to what extent pharmaceutical companies are able to secure the market share of their products. We focus on the over-the-counter analgesics market as to itshomogeneous properties and substantial group of consumers. A regression on brand share has been performed with the number of years a company was allowed to sell in a monopoly position and total yearly lobbying expenditures as explaining variables. The company’s total assets and its market
capitalization have been added as control variables. We find a highly significant positive effect of distributing years under exclusivity protection on brand share. We do not find evidence that implies a significant effect of total lobbying expenditures on the number of drugs sold relative to the total market sales.
This document is written by Milan van Aagten who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been
used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Table of Contents
Introduction 1. Welcome to the Pharmaceutical Industry 2. Literature review 2.1 Lobbying and political power 2.2 Advertisement of OTC medicine 2.3 Market share under generic entry 2.4 Physician-industry involvement 3. Econometric analysis Data 3.1 Defining the market 3.2 Dependent variable 3.3 Explanatory variables 3.3.1 Operating years in exclusivity 3.3.2 Estimating market dominance: Kwoka’s Dominance Index 3.3.3 Lobbying expenditures 3.4 Control variables 3.4.1 Direct-to-customer advertisement 3.4.2 Firm size indicators 3.5 Results Method 3.6 The model 3.7 Panel data specifications 3.7.1 The individual time series 3.7.2 Fixed effects estimation 3.8 Post-regression analysis 3.8.1 Linearity 3.8.2 White’s test for heteroskedasticity 3.8.3 Jarque-Bera test for normality 3.8.4 VIF on multicollinearity 3.8.5 Wooldridge-Drukker test for autocorrelation 4. Concluding remarks 4.1 In retrospect 4.2 Further research opportunities Referencing AppendixIntroduction
1.
Welcome to the Pharmaceutical Industry
The world of pharmaceuticals endures rapidly expanding pressure as time goes by. This is for a great extent due to industry dynamics and self-induced action although it is in conjunction with an invariant legal and economic environment in which the industry operates. With media outbursts as of Martin Shkreli’s 700% price hike, initiated by his company Turing Pharmaceuticals, of Daraprim™, a drug used for the treatment of aids complications. More recently Mylan sparked outrage with regard to its EpiPen™, following a 500 US dollar increase per package in just seven years. Mylan CEO Heather Bresch claims to be a: ‘’victim of a deficient health care system as intermediaries such as wholesalers and pharmacy benefit managers add to the ultimate cost resulting in a higher pay-check for patients’’(Mangan, D. CNBC, 2016). Both CEO’s argue to offer discounts to selected customers. This gesture however will be insufficient as higher drug prices ramp up insurance premiums as a secondary result . Concerns like these do not only raise questions regarding the motives behind price increases, it tests our judgment on how it is in any way justifiable. Including aforementioned, the American pharmaceutical industry is rich on scandals as corporate rap-sheets are pervasive enough to cover multiple decades. All of which have challenged the general consensus concerning how and if giant pharmaceutical companies should exert the amount of power they have grown to possess. Ongoing outrageous price hikes like Mylan’s will continue to ignite discussion regarding the ethical environment the industry is engaged in. It is evident to discuss both the controversy and academic gravity of the issue and how it is supported by existing literature., especially when taking into account that the negative effects of industry power have not thoroughly been exposed nor with sufficient backup.As no US law exists that protects patients or consumers from unreasonable price hikes by pharmaceutical firms, it might be helpful to observe the industry dynamics more closely. In the last decades the industry has been transformed drastically from merely a manufacturer of chemicals to a research oriented sector that contributes greatly to health care technology. The origin of success in the industry in generating a stream of new drugs with important therapeutic benefits has involved the industry in intense public policy debates on areas considering: research financing, the veracity of claims for its products, the prices charged for them (moreover who pays those charges) and the socially optimal degree of patent protection. The debate is mainly revolved around the welfare economic issue of the ongoing trade-off between promoting innovative effort and securing competitive market outcomes. The R&D sector of the industry is heavily reliant on patent protection measures . However while the monopoly induced profits from sales in exclusivity serve as a warrant for initial investment, they inhibit competition and thus increase deadweight loss in the market. Health care agents have continuously argued to be forced to prioritize and secure their R&D possibilities being victim of fast growing costs of drug innovation, thus claiming to righteously steer up their prices. It is generally presumed that drastic price hikes like aforementioned are labeled as unreasonable. It becomes therefore necessary to elaborate on the cases in which substantial price hikes are indeed justifiable. Scherer, (2001) shows that pharmaceutical companies are economically bound to invest vast amounts of capital in the research which is subordinate to invent new or higher quality medicine. With costs of developing new drugs having increased to 2.6 billion dollars in 2014 (PrMA Profile of Biopharmaceutical industry, 2015), pharmaceutical companies collectively claim to be forced to raise list prices.
Not to mention the investment expenditures in R&D conflict heavily with the lack of a corresponding increase in the number of medicine being approved by the U.S. Food & Drug Administration (Pammolli,, Magazzini, Riccaboni, 2011), which indicates of a more costly therapeutic innovation process. Nonetheless cases as that of Mylan do normally not revolve around new treatment methods. The controversy usually is not about medicine with groundbreaking therapeutic properties. It is silently assumed that despite these immense R&D costs there still is an outrageous amount of incoming cash flows left not defined to be used for operational factors only. My aim is to find a clarification of how big pharma-firms secure their brand share, even after competition raised by generic entrants . I am inspired to research how this directly affects us as patients. I will therefore subdivide the main question into two: • What impact does market power or a temporal privilege providing market power in the form of a patent have on brand share? • To what extent is a company able to affect regulation in its own advantage with monetary means? I will focus my research on analgesics that are sold over the counter1 (OTC). What makes this market especially interesting to study is the gap between the perceived distinction between products and the actual absence of differentiation due to the almost homogeneous nature of the products. The working substance differs solely between ibuprofen and acetaminophen (Table 1, Appendix). This rules out any quality based preferences a consumer can have. One of the core controversy sparks revolving around pharmaceutical companies is their increasing political influence (Henry, Lexchin, 2002). Especially around the affects on a company’s relative market power and ultimately the number of products they sell, to us. Possible outcomes of this research may allow us to speculate on often made substantial statements including; the deficiency of the US healthcare system, a contaminated pharmaceutical industry or neglecting corporate regulation.
1 As stated by the U.S. Food & Drug administration: ‘’OTC drugs are drugs that have been found to be safe and appropriate for use without the supervision of a healthcare professional such as a physician, and they can be purchased by consumers without a prescription.’’
2.
Literature review
The pharmaceutical industry has continuously been identified as an attractive area to investigate as it is subject to an abundance of scandals. Next to popular media being magnetized towards the industry’s latest movements, it attracts vast amounts of academics as the need for well-founded back up in news reports is higher than ever. The controversy and lack of trust revolving around the industry can only be at place if hard evidence is shown that conflicts with the arguments given for their actions. Attempts of dissecting the industry to allow for the construction of clear correlations are numerous and each shines light on a different element of the economics of health care. All have the aim however to ultimately provide us with some explanation on how the industry behaves. Existing literature on pharmaceuticals is segregated into three main fields: 1. The effects of direct-to-customer advertisement on OTC drugs 2. Physician involvement with pharmaceutical companies 3. Patent protection and the influence on brand share dynamics and generic competition All three of them will be elaborated on. Advertisement and more specifically the promotion of any therapeutic necessities has always been a classic controversial topic. The extent to which its competitive effect reaches has important implications for legislation around advertising, antitrust litigations and other legal procedures. In the 1999 paper by John Rizzo on antihypertensive medication, he finds that product detailing2 systematically lowers the price elasticity of demand, handing an explanation as to why pharmaceutical companies continuously raise prices with such ease. Contrary to this, Rizzo finds that the price the firm sets is of negative and highly significant effect on its sales, which thus concludes the disproportionally high compensating power of advertisement. He furthermore found that generic entry strongly contributes to this price sensitivity. Causes given in the paper all link to increased brand loyalty resulting from drug detailing. This is shown however to have detrimental effects on price competition in the market. These findings are consistent with Hurwitz and Caves (1998), who find evidence of the inhibition of entry into the market by advertising. They state that trademark holders’ sales- 2 Definition: Industry explaining of drug properties to a medical agent, perceived as equivalent to product marketing (Rizzo, 1999).promotion outlays do preserve their shares against incursion by generic entrants. Correspondingly their results show a diminished effect on brand share by incoming competition, after intense comparative advertising. Similarly to Rizzo they stress the industry’s high brand loyalty. As the vast majority of the OTC drugs are homogenous in nature or have similar therapeutic properties it is an attractive market to delve into. Anderson, Cilibertoz and Liaukonyte (2008) argue that the gap between actual and perceived discrepancy of in particular OTC analgesics is generated by advertising. The OTC analgesics market is characterized by substantial advertising expenditures by the market’s leading brands. Advertising-to-sales ratios range anywhere from 20-50%, being more than a sevenfold of the pharmaceutical industry’s average. The paper shows that consumers form their brand preference based on anything but the working substance the products possesses. All aforementioned conclude advertisement effects cannot be excluded from this research. Another focus point of existing literature on the health care system is industry involvement with health care agents, mostly being physicians. Chren and Landefeld (1982) found that physician interaction with drug companies, affected physician requests for drugs to be added to the hospital formulary. However more than half of the new medicine requested offered little to no therapeutic advantage over comparable drugs already on the formulary. Physicians are offered compensations ranging from gifts and industry paid meals to samples and even funding for travel and educational symposia. The attitudes about these expenses are divided and contradictory. As much as 85% of medical students believe it is improper for politicians to accept a gift, whereas only 46% found it improper for themselves to accept a gift of similar value from a pharmaceutical company (Wazana, 2000). Perhaps the concern is related to the potential consequences regarding costly prescription drugs or negative outcomes for health in general. The consensus about this topic among future physicians may indicate of a far stretching inherent urge for change. Another element that is essential of the industry is its historic patent protection policy. The Drug Price Competition and Patent Term Restoration Act3 ( also known as the Waxman-Hatch Act of 1984) was a major change in public patent policy. It served as a compromise between branded and generic producers and set a minimum for the effective period in which a drug developer can market exclusively. On the other hand it reduced the burden for generic entrants to gain government approval with respect to inordinate testing. Nonetheless, as patent protection has always been defined as a tradeoff between competition and innovation, Boldrin 3 http://www.fda.gov/newsevents/testimony/ucm115033.htm
and Levine (2013) find no empirical evidence to show that patents serve to increase innovation or productivity. Even when identifying the number of patents awarded with innovation they conclude no significant correlation. The main focus point in existent literature with respect to patent protection is the effects of expiration. The secondary effect, namely the entry of generic competition, has likewise been subject to a high level of investigation. Statman (1981) finds evidence suggesting that drugs protected by patent are able to hold on to their market positions beyond expiration, although in general generic and other name brands gradually capture an increasing share of each drug's market. Statman also states that generic competition in the hospital market is more vigorous than in the drugstore market, despite the relatively smaller size of the hospital market. This difference presumably is due to a greater incentive to economize or simply due to greater medicine knowledge. I would like to stress however there is a prominent lack of literature that magnifies on company efforts to gain power on areas outside of that of economical or market related power. The pharmaceutical industry sets high stakes at lobbying, spending more than 7 million US$ yearly on average4. The question concerning to what extent big pharma-firms exert the economic power they possess in affecting regulations, is one that will be carefully focused on in this paper. Withal I have still to encounter literature that discusses the number of years a company was able to distribute and sell its product under patent protection or has adopted it as explanatory variable for corporate performance. The temporary monopoly position patents offer is assumed to have long-lasting effects on brand loyalty and ultimately market share. My aim in this research is not only to address the explanatory variables of quality that is profoundly proven in existing literature, moreover it is to include influence coming from directions that have until now not been looked into.
4 Source: https://www.opensecrets.org/lobby
3.
Econometric Analysis
The following section will elaborate on the data that is consulted to analyze the effects of market dominance and socio-economic power on brand share. The methodology will be discussed including a thorough description of the empirical model and its validity. To conclude a subsection is devoted to give an extensive view on the regression results. Data 3.1. Defining the market Defining the market correctly is crucial in this research, as I am circumnavigating an important component of industry influence on health care, namely the direct physician-to-company contact. As mentioned before, physician involvement with pharmaceutical firms has been extensively investigated, showing a clear positive effect on the quantity of drugs that is added to the hospital formulary (Chren, Landefeld, 2000). Yet when focusing on over-the-counter medicine and its demand dynamics, any non-monetary compensation or symposia aimed at medical agents would have negligible effects, as the physician is highly limited in his attempts to affect the patient's preference on drug brands. Patients do not normally ask for their physicians’ advice regarding relatively harmless drugs bought over the counter. This offers a solution to the inaccessible data about physician involvement with pharmaceutical companies. Analyzing prescription drugs and their power in the market without including industry-physician contact would give heavily biased results. This is contrary to assessing OTC-drugs without bias when omitting the same variable. The sub-market of analgesics is particularly interesting due to the significant discrepancy between true differentiation of almost homogeneous medical products and the perceived differentiation among different brands. This is presumably originated in brand image differences created by drug advertisement. 3.2- Dependent variable Analyzing the competitive position of a specific drug asks us to apply the most accurate measure. An accurate representation of a product’s relative position in the market is its quantity sold, which directs us to the brand share per medicine as dependent variable. Data on this is retrieved from Passport database, consisting of time-series data on OTC drug brand shares from 1998 to 2016. Withal, the concept of this thesis arose from the controversy linked to price hikes in the pharmaceutical industry without any obvious economical consequence. This economical consequence is usually initiated by lower medicine sales or lower quantity demanded. This lays the foundation of defining the dependent variable as brand share, which isdefined as the relative sales of the medicine against the total analgesics sales. 3.3 Explanatory variables 3.3.1 Operating years in exclusivity5 or under patent6 Both patents and exclusivity are bound to a trade-off between innovation and competition. Offering a monopoly position for a limited time guarantees profits that warrant the manufacturer’s initial investment. This significantly decreases innovation costs while on the other hand negatively affecting competition in the market. Statman (1981) showed that drugs protected by patent are able to hold on to their market position beyond expiration dates although generics gradually capture an increasing market share. These findings are consistent with Frank & Salkever’s paper (1991) who finds a decline in price elasticity of brand name medicine demand stemming from market entry due to patent loss, indicating a diminished threat to brand name drug share. Existing theory implies significant effects of years a company has marketed its drug in exclusivity7 on (post-generic entry) market share, validating the variable’s explanatory function. 3.3.2 Estimating market dominance: Kwoka’s Dominance Index Appointing a measure for market power has been a comprehensive task. In the Glossary of Industrial Organization Economics and Competition Law (1993) market power is defined as: ‘’The ability of a firm (or group of firms) to raise and maintain price above the level that would prevail under competition.’’. Market power in this analysis will be represented by Kwoka’s Dominance Index: 5 ‘’Exclusivity refers to certain delays and prohibitions on approval of competitor drugs available and can run concurrently with a patent or not. It was designed to promote a balance between new drug innovation and generic drug competition.’’ 6 ‘’A patent is a property right issued by the United States Patent and Trademark Office (USPTO) to an inventor “to exclude others from making, using, offering for sale or importing the invention into the United States” for a limited time, in exchange for public disclosure of the invention when the patent is granted. Generally, the term of a new patent is 20 years from the date on which the application for the patent was filed in the United States. ‘’(5&6 both retrieved from : http://www.fda.gov/downloads/drugs/developmentapprovalprocess/smallbusinessassistance/ucm447307.pdf) 7 Specification of the variable: The year the patent has been issued or in which exclusivity has been given accounts for the starting year for the variable. The first year in which the patent is expired and thus is of no force along with absence of any exclusivity entitlement accounts for the closing year.
It emphasizes the gap between successive firms when they are ranked by size. The values of this measure range from 1 to 0, with the former value indicating a monopolistic market. Conversely, the closer to zero the measure is, the lower is the power of any single company (J. E. Kwoka, 1977). It is evident the index investigates on the market level and is unable to discuss company specific traits8. Thus an external factor, namely the competitive environment of the company, will be used rather a company based variable. KDI is therefore constant with respect to drug, though it remains time variant. 3.3.3 Lobbying expenditures My intention is to class companies by their national influence on regulations and by the level of protection they endure from governments or other interest groups such as trade unions. Their specific lobbying expenditures could be a precise indicator of this ‘’socio-economic weight.’’ The assumption is that this ultimately leads to brand loyalty, even after generic entry. Opensecrets.org provides an extensive database with annual lobbying spending per company, The validity of the data pool is confirmed by even providing specific lobbying reports sorted by name of the lobbyist and issue that is lobbied for or against. It is assumed efforts to gain influence on regulation take a substantial amount of time to be enforced. Therefore a 1-year lagged value of total lobbying expenditures is used in the regression to explain brand share. 3.4 Control variables 3.4.1 Direct-to-customer9 advertisement The most prominent omitted variable bias will be resulting from ignoring the immense advertising spending by pharmaceutical companies. The OTC-analgesics market is characterized by advertising expenditures comparable to their R&D spending when it comes to the top brands. Hurwitz and Caves (1998) find evidence that advertising inhibits entry into the market. They found trademark holders’ sales-promotion outlays do preserve their shares against incursion by generic entrants, which justifies advertisement spending as an independent variable. Datasets of a sufficient size however, are ought to be purchased which limits the data that is attainable. To solve I use an 2001-2005 average given in the Anderson, Cilibertoz and
8 I have considered taking a unit variable for market dominance being equal to 1 when market share would exceed the 40% (Definition retrieved from: https://stats.oecd.org/glossary/detail.asp?ID=3199). However the OTC-analgesics market showed to be highly competitive with market shares not exceeding the 20% benchmark.
9 While the advertising numbers include expenditures on multiple media, almost all of the advertising budgets were spent on broadcast television advertising, including network, cable, and spot TV (Anderson, Ciliberto, Liaukonyte, 2008).
Liaukonyte paper on OTC-analgesics advertisement in 2008. This average represents the share of total advertising expenditures by the top painkiller brands. I will use this share as a constant by lack of superior data, which presumably will account for the grand part of the omitted variable bias. Anderson (2008) furthermore exposes total advertisement spending to sales ratio, which would be a poorer representation assuming the share of voice10 to be linearly affected by advertisement spending. 3.4.2 Total Assets & Market Capitalization Other factors of influence on market share I accounted for are company size and correspondingly its market value. Yearly data is retrieved from Excel’s integrated DataStream for 1998 till 2016 with US dollars as valuta. 3.5 Results The results of the individual t-tests shown in table 5 show a clear positive correlation between the years a company’s product was protected from competition and its present market share. Furthermore interesting to notice is that none of the control variables actually contribute to the accuracy of the model. Not only lobbying expenditures experiences increased significance, Kwoka’s dominance index likewise seems to better explain brand share when the control variables are omitted, though at a 10% significance level. We therefore conclude that no evidence was found to suggest inequality in the market impacts a drug’s market share.
10 Share of voice is defined as the ‘’degree to which people see a company’s advertisements, etc. in a particular market, compared to those of competitors’’. ‘share of voice. (n.d.) Cambridge University Press. (2008) Retrieved from
There is no evidence of significant influence of a higher Kwoka’s Dominance Index on pharmaceutical sales. This can either hint on a deficient variable to measure market dominance or of more conclusively, no significant effect of an unequal market on brand share. Reasons for this may include that the emergence of a competitive environment is a comprehensive and lengthy process, which is assumed to be highly time-invariant. It presumably fluctuates most during the years following patent expiration, years in which generic entry rapidly increases. Combinations of lagged values could therefore be of greater value , however this requires their joint significance to be tested. The increase in the model’s explaining power after removing one of the control variables that indicate firm-size, may be resulting from multicollinearity between market capitalization and total assets. As both imply the relative size and magnitude of a company, omitting one does indeed improve the model.
Method
3.6 The Model The setup of the regression can be described is as follows. Panel data is analyzed for years 1998 – 2016. 𝐵𝑟𝑎𝑛𝑑𝑠ℎ𝑎𝑟𝑒!" = 𝐵!𝐸𝑥𝑐𝑙𝑦𝑒𝑎𝑟𝑠!+ 𝐵!𝐿𝑜𝑏𝐸𝑥𝑝!" + 𝐵!𝐾𝐷𝐼!+ 𝐵!𝑀𝑎𝑟𝑘𝑒𝑡𝐶𝑎𝑝!"+ 𝐵!𝑇𝑜𝑡𝑎𝑙𝐴𝑠𝑠𝑒𝑡𝑠!" + 𝐵!𝐴𝑑𝑣𝑆ℎ𝑎𝑟𝑒!+ 𝑣!" : Idiosyncratic error113.7
Panel data specifications
3.7.1 The individual time series The fact that we are dealing with yearly time series data raises some essential questions that need to be answered before analyzing linear relationships between the time series. Presence of stationarity of the time series is what requires to be examined first. We can encounter the risk of spurious regression if the series happen to be non-stationary12. Spurious regression indicates a professedly strong inter-time series relationship, which leads to the estimators being inconsistent. Testing for stationarity is commonly executed by a Dickey-Fuller test. However due to our sample size that has a maximum of 95 observations the significance of the test is inherently decreased, thus we can only assume stationarity. To avoid inconsistent estimates based on this assumption, we consult Granger and Newbold (1974) in that the R2 of the model cannot be greater than the Durbin-Watson statistic, otherwise being indicative of a spurious regression. The Durbin-Watson statistic is not applicable on panel-data therefore placing a statistical analysis of stationarity out of reach within the scope of the present analysis. If we focus on the properties of brand share we can reason that it is highly dependent on other company behavior and their corresponding market shares. Considering the significant effect of advertisement influence on brand loyalty and ultimately number of products sold (Rizzo 1999) we can argue theoretically that brand share does not have a unit root, as a unit root would imply a weak or strongly lagged effect of advertisement. In his research Rizzo avoids using lagged values of marketing expenditures implying a direct effect that is not lagged. 11 Time invariant & drug invariant error terms omitted. This is elaborated on in ‘’Panel data implications’’ 12 http://sergei-sarkissian.com/papers/spurious.pdf-5 0 5 R esi du al s 0 5 10 15 Fitted values 3.7.2 Fixed effects estimation Primarily we take into account that pooled (per drug) OLS will not be applicable here due to unobserved heterogeneity, conflicting with the Gauss-Markov assumption of a zero conditioned mean for the error term13. A Hausman-Test shows the initial null hypothesis is soundly rejected, which stated the individual-level effects are adequately modeled by a random-effects model. The covariance between the time-consistent error and explanatory variable therefore does not hold to be equal to 0, justifying a fixed effects estimation. Nonetheless as estimating with fixed effects removes the effect of time-invariant characteristics, it will subsequently omit any time-constant variables. The years the firm has sold the drug in exclusivity as well as the average share of advertisement spending are both time-constant resulting in fixed effects estimation being unsuitable. Settling for this issue requires to add yearly dummy variables combined with dummies for each pharmaceutical and excluding fixed effects estimators. This will prevent any time-constant variable being omitted. 3.8 Post-regression analysis 3.8.1 Linearity
One of the Gauss Markov assumptions states the dependent variable, in our case market share, requires to be a linear function of the set of explanatory variables. If we plot the residuals against the predicted values it becomes clear whether or not we deal with nonlinearity. The points should be distributed around the horizontal line in a symmetrical manner. As can be seen in Figure 2 the points
start symmetrically distributed however as the values increase, more outliers become visible. This violation of linearity can lead to miss-specified determinants. However interpreting this distribution as nonlinear would be impetuous, taking into account the limited range of the sample size. 13 Odell, P. L. (1983). Gauss–Markov theorem. Encyclopedia of statistical sciences. Figure 1. Residuals versus fitted values
Figure 2. Normal quantile plot of residuals 3.8.2 Heteroskedasticity A White’s test has been applied for heteroskedasticity, which resulted in rejecting the null hypothesis stating that the variance of our errors given the explanatory variables is equal to 0. We can assume heteroskedasticity to be present and have accounted for it by adding robust error terms to our regression. 3.8.3 Normality of the residuals Violations of normality can create problems for determining whether coefficients are significantly different from zero. A skewed error distribution is ought to be excluded stressing the absence of any large outliers. To give an indication of the error term distribution a normal quartile plot of the residuals is shown in Figure 2. As the plot is sensitive to deviations from normality in the tails we can conclude there is no evidence of non-normality. This agrees clearly with what is revealed in the Kernel Density Estimate (Figure 3). Nonetheless, considering the relatively small sample size a Jarque-Bera test is performed to fully secure the absence non-normality. The outcome did not allow for rejecting the null-hypothesis of having normally distributed error terms.
3.8.4 Multicollinearity Variance Inflation Factors (VIF) have been used to test a correlation between any of our explanatory variables. Multicollinearity will manifest in our model by inflated standard errors, which damages the significance of the estimators. Applying VIF resulted in absence of any collinearity, as no value that was calculated exceeds 4 (Table 3). These findings are consistent with the Pearson values exposed in the correlation matrix (see Appendix, table 4).
3.8.5 Autocorrelation To test if there is a relationship between the error terms I performed a Wooldridge-Drukker test. The reasoning behind preferring the Wooldridge test over the commonly used Durbin-Watson test statistic is its simplicity with respect to panel model testing along with the absence of large tables providing the critical values of both the cross-section dimension (i=1,..., N) and the time-dimension (i=1,...,T) (Wooldridge, 2010). The test indicated we could not reject the null-hypothesis suggesting there is indeed autocorrelation. The outcome raised from performing the Wooldridge test can in principle originate from three complications: 1. The missing of a decisive explanatory variable, causing omitted variable bias 2. A functionally miss-specified the model 3. A measurement error in one or multiple explanatory variables From these, a measurement error of an independent variable is most plausible. For advertisement spending, which is proven to be of high and significant effect on OTC-pharmaceutical brand share (Anderson, Cilibertoz, Liaukonyte, 2008), the variable used is to a great extent unspecified. Advertising can be subdivided to comparative or detailing and auditive or visual, all having different impact levels. Additionally a constant value is used instead of a linear estimation, as due to the limited data used. 4. Concluding remarks 4.1 In retrospect The results show a clear relation between the years the company could distribute its product under protective measures and the brand’s share of the market. Creating a monopoly environment, even for a limited time is a privilege to account for in the pharmaceutical industry. An explanation for this can be that when it comes to health, people tend to compose their brand loyalty in an accelerated fashion. A complementary risk-averse attitude towards trying new drugs, while being content with an alternative, is likely to exist. Focusing on lobbying expenditures we can conclude there is not enough statistical evidence found to claim they have significant positive effects on a products sales. Several reasons for this can be acclaimed. To start lobbying efforts in the model are defined at the company level. As many players in the pharmaceutical market also have investments and stakes in other
industries , it can differ from year to year what percentage of lobbying effort is specifically aimed at OTC drug reregulation. The second explanation may be rooted in the lagged effect on regulation influence attempts. Creating new legal structures takes a tremendous amount of time, not even considering its implementation. If the most prominent pharma-firms communicate their ideals it will require revising of existing and future laws which furthermore lengthens the process. Therefore using a 1-year lag might be suboptimal, although our dataset did not allow for a great number of lags. A third explanation contains the more complicated web of regulation around prescription drugs, relative to that around medicine bought over the counter. This may indicate of that a great part of lobbying is targeted at prescription drug-related issues. The share of the total market advertisement spending has not been found to have a significant effect on a brand’s market share. This may show the complicated manner of defining and subdividing advertisement. Anderson, Cilibertoz and Liaukonyte (2008) state that whether drug promotion is of comparable nature or not has great positive influence in generic entry inhibition, stressing the importance of classifying advertisement. Furthermore advertisement expenditures are assumed to be linearly correlated to company benefits, or if advertisement spending correlates to a higher share of voice, the relationship may be hyperbolic (Rizzo, 1999). In this research a constant is used which is highly likely to negatively impact the variable’s explanatory significance. As Hurwitz and Caves (1998) emphasize the high level of advertising impact on OTC drug sales, a not ideally specified regressor can result in serial correlation, which we indeed found evidence of using variance inflated factors. Remarkably there was no evidence of multicollinearity in the model. The total amount of assets a firm has next to its total market capitalization are both indicators of firm size, which suggests of a degree of correlation. A possible explanation for this is that assets are balanced by the burden of liabilities a company carries, while its value calculated from stock price may be a more honest market representation of its value. 4.2 Further Research Opportunities A great part of the initial motivation for this paper lies in the direct effect on us as patients. Questioning how the growing political entanglement of the pharmaceutical industry together with frequent outrageous price hikes correlate to consumer healthcare and consumer behavior has been key to the construction of this paper. Examining the quantity that is purchased of over-the-counter pharmaceuticals can only partly cover this matter. A more accurate and authentic measure for industry socioeconomic power would the price elasticity of demand, in particular
that of prescription drug. Although existing literature spreads from advertising influences on OTC market share to the relation of physician-company contact with drugs added to the hospital formulary, I have yet to encounter research which focusses on the effects of regulation and efforts to affect the legal environment it forms. Ideally a time-series investigation would be executed including data on: ● advertisement, share of voice or expenditures ● patent protection & exclusivity ● generic entry & market share changes ● efforts to gain socio-economic power, both lobbying investments and involvement of industry with health care agents ● price elasticities of demand This combined with a dataset that reaches far before and after generic entry would give interesting insights on the nature of this complicated industry.
Reference list Anderson, S. P., Ciliberto, F., & Liaukonyte, J. (2008). Getting into your head (ache): Advertising content for OTC analgesics. Marketing Science Institute Working Paper. Avorn, J., Chen, M., & Hartley, R. (1982). Scientific versus commercial sources of influence on the prescribing behavior of physicians. The American journal of medicine, 73(1), 4-8. Boldrin, M., & Levine, D. K. (2013). The case against patents. The journal of economic perspectives, 27(1), 3-22. Kwoka, J. E. (1977). Large firm dominance and price-cost margins in manufacturing industries. Southern economic journal, 183-189. Frank, R. G., & Salkever, D. S. (1991). Pricing, patent loss and the market for pharmaceuticals (No. w3803). National Bureau of Economic Research. Granger, C. W., & Newbold, P. (1974). Spurious regressions in econometrics. Journal of econometrics, 2(2), 111-120. Henry, D., & Lexchin, J. (2002). The pharmaceutical industry as a medicines provider. The Lancet, 360(9345), 1590-1595. Hurwitz, M. A., & Caves, R. E. (1988). Persuasion or information? Promotion and the shares of brand name and generic pharmaceuticals. The journal of law and economics, 31(2), 299-320.
Mangan, D. (2016, August 25). ‘’Mylan CEO Bresch: 'No one's more frustrated than me' about EpiPen price furor.’’ CNBC. Retrieved from: http://www.cnbc.com/2016/08/25/mylan-expands-epipen-cost-cutting-programs-after-charges-of-price-gouging.html Odell, P. L. (1983). Gauss–Markov theorem. Encyclopedia of statistical sciences. Pammolli, F., Magazzini, L., & Riccaboni, M. (2011). The productivity crisis in pharmaceutical R&D. Nature reviews Drug discovery, 10(6), 428-438.) Rizzo, J. A. (1999). Advertising and competition in the ethical pharmaceutical industry: The case of antihypertensive drugs. The Journal of Law and Economics, 42(1), 89-116
Scherer, F. M. (2001). The link between gross profitability and pharmaceutical R&D spending. Health Affairs, 20(5), 216-220. Statman, M. (1981). The effect of patent expiration on the market position of drugs. Managerial and Decision Economics, 2(2), 61-66. Wazana, A. (2000). Physicians and the pharmaceutical industry: is a gift ever just a gift?. Jama, 283(3), 373-380. Wooldridge, J. M. (2010). Econometric analysis of cross section and panel data. MIT press.
Appendix
Brand Shares (Global - Historical Owner) | Historical | Retail Value RSP | % breakdown
Geographies USA USA USA USA USA
Categories Analgesics Analgesics Analgesics Analgesics Analgesics
Brand Tylenol Advil Aleve Motrin Bayer Aspirin
Effective substance Acetaminophen Ibuprofen Naproxen Ibuprofen Aspirin
Company name (GBO) Johnson & Johnson Inc Pfizer Inc Bayer AG Johnson & Johnson Inc Bayer AG
Years in exclusivity 8 10 6 2 2 1998 11.80 14.60 6.60 4.60 5.20 1999 11.60 14.80 6.90 4.70 5.10 2000 10.90 15.10 7.90 5.00 4.80 2001 11.20 13.90 8.20 5.40 4.90 2002 13.00 13.40 8.40 5.30 4.80 2003 14.30 12.80 8.30 5.60 4.70 2004 15.60 12.50 8.10 5.80 4.10 2005 16.30 14.10 7.90 6.20 4.20 2006 16.50 14.30 7.80 6.10 4.30 2007 16.70 13.00 7.50 6.30 4.20 2008 16.80 12.00 7.10 6.40 4.70 2009 16.50 12.40 7.00 6.60 4.60 2010 9.80 14.00 7.80 3.20 4.90 2011 7.20 15.00 8.10 1.00 5.40 2012 5.10 15.90 8.70 1.40 5.70 2013 6.70 14.80 8.50 2.50 5.40 2014 7.80 14.10 8.60 2.90 5.30 2015 8.60 13.90 8.10 3.40 5.10 2016 8.90 14.60 7.30 4.00 5.10
Table 1. Analgesics Legenda
-5 0 5 R esi du al s 0 5 10 15 Fitted values
Figure 4. Residuals versus fitted values
Figure 3. Normal quantile plot of residuals
