This item is only available as the following downloads:
The Anticompetitive Effects of The Apple E book Price Fixing Conspiracy William Giroux and Dr. Roger Blair Department of Economics, University of Florida On July 10th, 2013, Apple was found guilty of conspiring with publishers Macmillan, HarperCollins, Hachette, Simon & Schuster and Penguin (the "defendant publishers") to raise the price of e books. Resembling a traditional hub and spoke conspiracy, Apple d rafted nearly identical agreements with each defendant publisher in order to achieve the conspiracy's ultimate goal of raising e book prices and ending Amazon's loss leader strategy of selling e books for $9.99. These Apple Agency Agreements transformed th e e book industry from a wholesale model to an agency model, thereby shifting control over retail e book pricing from the retailers to the publisher defendants. The Apple Agency Agreements contained three essential elements sufficient to achieve this: a mo st favored nation clause, e book price tiers set by Apple and a stipulation that Apple would receive thirty percent commission on each e book sold through the iBookstore. The Apple Agency Agreements essentially forced the defendant publishers to shift all of their retailers to an agency model, ensured e books would be priced as high the price tiers allowed and eliminated retail price competition to Apple's benefit. I believe this conspiracy led to substantial increases in the price of e books and reductions in unit sales, and therefore caused significant harm to consumers. I will use regression analyses and publisher specific sales data to show the anticompetitive effects of this conspiracy and prove that consumers sustai ned significant harm. I NTRODUCTION Motivations for Collusion Before April 2010, e book retail pricing was determined through a wholesale model, whereby the publisher defendants would sell e books to retailers, who then determined the retail price charged to consumers. Disaffe ction amongst the publisher defendants began to form when Amazon, the dominant e book retailer at the time, began selling New York Times Bestsellers and New Release e books for $9.99, well below the wholesale price charged by the publishers. 1 The publishe rs feared low e book prices would diminish the perceived value of books in general, creating a new low price standard consumers would come to expect for all books, both print and digital. The publishers worried this would affect the much more profitable pr int book markets by reducing their ability to charge substantially higher prices for hardcover titles, which threatened the viability of their brick and mortar stores. The publisher defendants also feared that Amazon would exercise its significant monopson y power and demand lower wholesale prices for e books. Finally, the conspiring publishers worried that Amazon would soon offer authors higher royalties 1 Amazon controlled 90% of e book market revenues during 2009 and 80% of market revenues in the first quarter of 2010. than the publishers currently provided, resulting in disintermediation. 2 Beginning in 2009, the pub lishers jointly executed a variety of strategies to pressure Amazon towards higher prices, including eliminating wholesale price discounts and windowing, all with little success. Obstructing the defendant publishers' effectiveness was the fear of retaliati on by Amazon in response to unilateral publisher action against the retail giant. As President and CEO of Simon & Schuster Carolyn Reidy noted, "without a critical mass behind us Amazon won't negotiate." 3 With the launch of the iPad drawing near, Apple n eeded to secure sufficient content for the iBookstore to certify the iPad's competitiveness as a legitimate e reader. With content acquisition in mind, Apple recognized it was the critical mass the publishers could unify around to end the $9.99 problem and began to take steps to ensure this unification. 4 Apple became the defendant publisher's information hub in December 2009, disseminating all things relevant to the conspiracy amongst the publishers. Every email, telephone call or meeting between Apple an d one publisher defendant was immediately relayed to the other co conspirators, making Apple the center of this emerging hub and spoke conspiracy. This free flow of 2 Apple Inc., 952 F. Supp. at ¦ 15 16. 3 Reidy, Carolyn, email message to Leslie Moonves, September 23, 2009. 4 Apple Inc., 952 F. Supp. at ¦ 29.
information gave the publishers confidence that this was a concerted effort and critically allayed their fears of retaliatory action by Amazon. These efforts culminated in the creation of the Apple Agency Agreements in mid January 2010, the mechanism that transformed the digital publishing industry and facilitated a conspiracy to charge supracom petitive e book prices. The Apple Agency Agreements The Apple Agency Agreements contained three essential components that in combination facilitated the conspiracy to raise e book prices: price tier caps, a fixed commission on e book sales and a most f avored nation clause. The Apple Agency Agreements stipulated that Apple and other retailers would act as the publishers' agents, thereby transferring control over retail pricing to the publishers. The agency agreements set price caps for e books sold throu gh the iBookstore and other retailers as a percentage of its corresponding hardcover list price. These price caps, set by Apple, ranged from $12.99 to $14.99 for New York Times Bestsellers and from $9.99 to $19.99 for all other New Releases. Backlist title s were not subject to a cap. In addition, the Apple Agency Agreements stipulated that retailers would earn a thirty percent commission on every e book sold, regardless of its retail price. Most importantly, the most favored nation clause required the publi shers to match in the iBookstore any lower retail price offered by any other retailer for a NYT Bestseller, New Release or Backlist e book. 5 The Apple Agency Agreements ensured two things: e books would be priced at their respective caps and the defendant publishers would force all of their retailers, including Amazon, to an agency model. Under the wholesale model, the defendant punishers usually received approximately $13 to $15 in revenue per e book and sometimes as much as $17.50. 6 Under the agency mode l, the defendant publishers would receive a maximum of seventy percent of $14.99 for each NYT Bestseller sold through the iBookstore, or $10.49. This is over $2 less than the $13 minimum they received at wholesale. Given these losses are qualitatively repr esentative of most if not all e books sold under agency, the conspiring publishers earned less revenue on each e book sold. Therefore, it was in the their best interest to price e books at their maximum caps in order to minimize losses. The publishers were willing to take these losses because as stated previously, they were focused on their long term interests of protecting hardcover markets, preventing wholesale price concessions and eliminating disintermediation. 5 Apple Inc., 952 F. Supp. at ¦ 52. 6 Id. at ¦ 33. Moreover, if the publishers permitted an y of their retailers to remain on the wholesale model and if those retailers set prices below the caps, then per the MFN clause, the publishers would have to match that lower price in the iBookstore. Because the publishers would be earning less revenue per e book when prices were at their maximum cap, allowing their retailers to maintain price control would risk even lower retail prices, leading to further profit erosion. In this regard, the MFN clause was essentially a financial penalty that forced the def endant publishers to convert all their retailers to agency to prevent any further reductions in earnings. 7 Thus, the Apple Agency Agreements effectively eliminated retail pricing competition for Apple, who was ensured e book s nearly industry wide would be priced at their ca ps. The evidence of this is clear. Shortly after the Apple Agency Agreements went into affect beginning on January 21st, 2010, Amazon was pressured onto agency by defendant publishers in early February 2010, as was Google that same January. Additionally, Table 1 below shows the percenta ge of the publisher's e books sold through Apple and Amazon that were priced at their caps five months after agency implementation. Table 1 shows HarperCollins, Macmillan and Penguin priced nearly all of their e books sold through Apple at their respective caps, with most others pricing near or substantially above 9 0% of their e books at the cap 7 Id. At ¦ 12
Table 1: Degree to Which Prices Were at the Applicable Price Cap Retailer Apple Amazon Publisher New Releases NYT Bestseller New Releases NYT Bestseller Hachette 96.3% 99.7% 89.9% 100.0% HarperCollins 90.0% 100.0% 84.6% 95.6% Macmillan 81.1% 100.0% 76.3% 98.7% Penguin 98.4% 100.0% 92.2% 99.3% Simon & Schuster 91.3% 97.9% 83.7% 90.1% Defendant Publishers 92.1% 99.4% 85.7% 96.8% Source : Gilbert Report, Table 4 at ¦ 48. METHODS The Regression Model To determine the price and output effects of the conspiracy, I utilized two expert reports prepared by Dr. Orley Ashenfelter and Dr. Richard Gilbert for the Department of Justice. 8 9 To measure the competitive effects of the conspiracy, Dr. Ashenfelter compared actual e book prices and unit sale s to what prices and unit sales would have been, but for the conspiracy. An analysis of difference in differences was used, whereby the change in average e book prices and units sold by the publisher control group, Random House, was subtracted from the cha nge in the publisher defendants' average prices and units sold in order to determine the impact of the conspiracy on prices and unit sales. Changes in price and unit sales of e books published by Random House, the largest of the big six publishers and havi ng not implemented agency until January 2011, reflects the market conditions that would have affected the publisher defendants, but for the conspiracy. This makes Random House a superior control group with which to calculate the effects of the conspiracy. The primary analysis utilizes three regression models over varying time horizons to determine the impact of the conspiracy. For the purposes of this discussion, the pre period is defined as the length of time before agency pricing was implemented, while the post period is defined as the length of time after agency pricing was implemented. The regression analyses assume that agency pricing began on April 1st, 2010. The data used to calculate price and output effects consisted of e book sales data supplied by retailers Amazon, Barnes & Noble and Apple. The results of these regressions are listed in tables 2 though 4 and display in log points the average change in aggregate e book price, units sold and 8 Ashenfelter Report 9 Gilbert Report revenue from the specified pre period to the post perio d for the conspiring publishers The tables also display the calculated price elasticity of demand over the time horizon tested and a T statistic greater than 1.96 shows statistical significance. Effects Specific to Each Defendant Publisher Dr. Richard Gilbert prepared a similar analysis. Dr. Gilbert computed average price changes for the defendant publishers' e books at Amazon and Barnes & Noble to isolate the price effects of the conspiracy. Per unit e book prices were calculated for February 2 010 and February 2011, and prices changes over these two one month periods represents the pr ice effects of the conspiracy. The analyses assume agency was implemented in early April 2010. Sources of data used to calculate the average change in price consist s of e book sales data for each publisher defendant at retailers Amazon and Barnes & Noble. The results of these analyses are displayed in table 5. RESULTS Prices Rose and Unit Sales Fell Tables 2 through 4 represent the results of Dr. Ashenfelter's re gression analyses on price, unit sales and revenues for the publisher defendants' Backlist, New Release and NYT Bestselling e books. Table 2 shows the results of the regression analysis using October 2009 March 2010 as the pre period and April 2010 Sep tember 2010 as the post period. The table indicates that e book prices of the conspiring publishers rose 16.8% on average and unit sales fell 14.5% on average. Both results are statistically significant.
Table 2 : Results of the Regression Analyses of E book Prices and Units Sold Using a Six Month Pre and Post Period Price Units Sold Revenue Effect of the Agency Model 0.155 0.157 0.002 Absolute value of the T statistic (90.57) (22.59) (0.24) Number of Titles 27,791 27,791 27,791 Total Number of Observations 406,440 406,440 406,440 Elasticity 1.01 Source : Ashenfelter Report, Table 1. Table 3 shows the results of the conspiracy using March 7 th 20 th 2010 as the pre period and April 11 th 24 th as the post period. Over this time period, prices rose 13.7% on average and unit sales fell 7.5% on average. Both results are statistically significant. Table 3: Results of the Regression Analyses of E book Prices and Units Sold Using a Two Week Pre and Post Period (Penguin Excluded) Price Units Sold Revenue Effect of the Agency Model 0.129 0.078 0.050 Absolute value of the T statistic (67.48) (8.76) (5.72) Number of Titles 24,496 24,496 24,496 Total Number of Observations 62,728 62,728 62,728 Elasticity 0.61 Source : Ashenfelter Report, Table 3. Note: Although Penguin executed its agency agreements at Apple and Barnes & Noble in early April of 2010, it did not execute agreements at Amazon until late May of 2010. Consequently, price and unit sales data for Penguin are omitted from this analysis. Table 4 shows the results of the conspiracy using February 2010 as the pre period and February 2011 as the post period. Under the longest time frame studied, prices rose 24.6% and unit sales fell 19. 3% on average. Both results are statistically significant. Table 4: Results of the Regression Analyses of E book Prices and Units Sold Using a One Year Pre and Post Period Price Units Sold Revenue Effect of the Agency Model 0.220 0.215 0.005 Absolute value of the T statistic (86.13) (19.41) (0.50) Number of Titles 34,939 34,939 34,939 Total Number of Observations 93,062 93,062 93,062 Elasticity 0.98 Source : Ashenfelter Report, Table 4. The results of the conspiracy are uniform regardless of which time horizon is selected. Prices rose, unit sales fell and consumers were harmed as a result. Notice that as the length of time the conspiracy was in place increases the effects of rising price s and falling output intensifies This ind icates that over time, the harm to consumers resulting from the conspiracy became increasingly more severe. Table 5 looks at average per unit price increases for the defendant publishers' e books from February 2010 to February 2011 at retailers Amazon and Barnes & Noble. The table quantifies the aggregate effect on e book prices resulting from the conspiracy for each publisher defendant, Random House and Non Major publishers. The aggregate changes in the price of NYT Bestsellers, New Releases and Backlist e books are also listed. Table 5: Summary of E book Price Increases At Amazon and Barnes & Noble by Defendant Publishers from February 2010 to February 2011 Amazon Average Per Unit Price Increases Publisher All e books New Releases NYT Bestseller Backlist Hachette 48.4% 36.8% 60.4% 53.4% HarperCollins 19.3% 18.5% 22.3% 28.8% Macmillan 8.8% 13.7% 30.8% 10.4% Penguin 21.5% 24.7% 36.2% 24.0% Simon & Schuster 21.2% 18.2% 26.1% 23.2% Defendant Publishers 23.9% 24.2% 40.4% 27.5% Random House 17.1% 5.0% 4.8% 19.2% Non Majors 20.0% 7.6% 0.6% 22.4% Barnes & Noble Average Per Unit Price Increases Publisher All e books New Releases NYT Bestseller Backlist Hachette 49.2% 40.5% 78.0% 44.9% HarperCollins 10.4% 11.6% 12.9% 5.5% Macmillan 4.0% 8.3% 15.3% 1.1% Penguin 18.5% 18.2% 38.6% 18.7% Simon & Schuster 19.7% 20.1% 26.9% 22.8% Defendant Publishers 19.3% 18.1% 48.6% 19.2% Random House 17.7% 11.9% 2.9% 16.3% Non Majors 30.5% 3.2% 7.0% 37.1% Source : Gilbert Report, Table 6 at ¦ 55. Table 5 shows that over this time horizon, e books sold by the conspiring publishers through Amazon and Barnes & Noble sustained substantial price increases. Collectively, the prices of the defendant publishers' e books rose by 23.9% at Amazon and 19.3% at Barnes & Noble. Focusing on specific categories of e books,
defendant publishers' NYT Bestseller prices rose 40.4% at Amazon and 48.6% at Barnes & Noble. Additionally, New Releases sustained price increases of 24.2% at Amazon and 18.1% at Barnes & Noble. Notice as well that e books published through Random House and Non Major publishers, who had not adopted the agency model over this period, experienced a decline in aggregate e book prices at both retailers. As explained previously, Random House is a sup erior model of the but for world and demonstrates the market conditions that would have affected industry e book prices absent the conspiracy. Thus, because e book prices at Random House declined from February 2010 to February 2011, industry wide e book pr ices would likely have fallen as well, but for the conspiracy. This indicates consumers paid a significant overcharge when purchasing the defendant publishers' e books. Dr. Gilbert also calculated short run changes in e book unit sales. Comparing unit sa les during the two week period ending March 20 th 2010 to the two week period ending April 24 th 2010, defendant publishers experienced a total decline in unit sales of 77,307 or 12.9% at retailers Amazon, Barnes & Noble and Apple. Over the same time perio d, e book unit sales of non conspiring publishers at these retailers rose by 58,980 units or 5.4%, leading to a net reduction in industry sales of approximately 18,327. 10 This net reduction of 18,327 represents consumers who were foreclosed from the e book market as a result of the conspiracy and corresponds to a direct loss in consumer welfare. As the regressions in tables 2 though 4 show, reductions in unit sales and inc reases in price intensified and the price elasticity of demand became more elastic over time, reaching a peak one year post agency. Given the net reduction of 18,327 units was calculated when price and output effects were minimal relative to their long run levels, it is reasonable to conclude that this net reduction in e book industry sales intensified over time, assuming the growth in the non conspiring publishers' unit sales continued to be less than the decline in the defendant publishers' unit sales. Th is indicates that net reductions in industry sales would be greater in the long run, corresponding to a larger number of foreclosed consumers. Harm to Consumers: Overcharges, Utility Diminutions and Market Foreclosures When prices rise and output falls simultaneously, there will be a loss in consumer welfare without any offsetting factors. It has been shown that as a direct result of the Apple e book price fixing conspiracy, e book 10 Gilbert Report, Figure 2 at ¦ 20 22. Note: Unit sales data does not include unit sales of Penguin. prices rose and unit sales fell. The anticompetitive harm to consumers re sulting from the conspiracy consists of consumers who sustained an overcharge, were forced to purchase a less desired e book and those foreclosed from the market entirely. 11 The overcharge represents the portion of consumer surplus transferred to the consp iracy as profit. This can be calculated as the difference in retail prices under wholesale and retail prices under agency, multiplied by the number of units sold during the conspiracy. Given that the conspiring publishers experienced a decline in unit sale s of approximately 20% from February 2010 to February 2011, nearly 80% of their pre conspiracy output was sold to consumers at agency prices. 12 If the relationship between price and units sold is linear, then the calculation of the overcharge is as follows: Overcharge = (P agency P wholesale ) x Q sold In figure 1 below, the overcharge is represented as: Overcharge = (P a P w ) x Q a This calculation must be performed for every Backlist, NYT Bestseller and New Release e book sold during the conspiracy. The s ummation of these calculations would equal the total overcharge sustained by consumers during conspiracy's damages period of April 2010 to May 2012. Similarly, many consumers sustained harm because they could not purchase the e book they truly desired fr om the conspiring publishers due to prohibitively expensive prices. These consumers were consequently forced to purchase a lower priced e book they did not truly prefer from a non conspiring publisher. This effect is much more difficult to quantify because the harm incurred would equal the utility obtained by consuming the e book desired less the utility obtained through the e book actually consumed, with a monetary value then being attached to this difference in utility. Such determinations are subject to considerable variability and estimates could vary significantly. As explained previously, declines in consumer welfare stem also from consumers who are foreclosed from the market entirely due to supracomeptitive retail prices exceeding their reservation price. Assuming the relationship between price and quantity is linear, this dead weight loss would be equal to: Dead Weight Loss = (dQ)(dP) 13 11 Apple Inc., 952 F. Supp. at ¦ 98. 12 Ashenfelter Report, Table 4. 13 Noll Report at ¦ 7.
where dQ and dP are the changes in unit sales and price during the conspiracy. A qualitative analysis of the an ticompetitive harm to consumers caused by this conspiracy can be shown graphically using a simple demand model, as shown in figure 1. In figure 1 below, the publisher defendants are initially operating at point B on the demand curve, producing Q W units of output at price P W Given this price and output, consumer surplus is equal to area 1/2(Q W )(C P W ), where C is some price resulting in zero output. Implementation of agency is represented in figure 1 by the rise in price from P W to P a and the fall in output from Q W to Q a or point A on the demand curve. Thus, the new level of consumer surplus is given by area 1/2(Q a )(C P a ). Given figure 1, the effects of the conspiracy are visually clear; when prices rise and output falls, there is a decline in consumer surplus. Although this model assumes a simple single tier price and output, this analysis would extend to every Backlist, New Release and NYT Bestseller sold by the defendant publishers during the conspiracy. All have been shown to have experienced similar price increases and output reductions, differing only in magnitude. Thus, the aggregate effect of the conspiracy is effe ctively illustrated in this model; prices rose, output fell and consumer welfare declined as a result. Figure 1: Dec r ease in Consumer Surplus CONCLUSION Both Dr. Gilbert and Dr. Ashenfelter have independently reached the same conclusions regarding the Apple e book price fixing conspiracy; e book prices rose and unit sales fell. Consumers sustained significant harm as a result of this conspiracy. To an extent, losses in consumer welfare were offset by the utility derived by the revolutionary new iPad and falling Kindle prices. However, it was established at trial that these factors did not lead to a net increase in consumer welfare nor were they found to have been causally related to the price fixing conspiracy. Thus, consumers sustained significant harm overall. Apple s participation as the center of this hub and spoke conspiracy was absolutely critical to its success. The legality of this conspiracy was subject to a per se analysis at trial, and Apple has subsequently appealed the decision claiming they are free of wrongdoing. Prior to Apple's involvement, the publishers began to conspire to end Amazon 's $9.99 pricing using strategies such as windowing in early 2009. When Apple drafted the Apple Agency Agreements, they were participating in an already existing horizontal price fixing conspiracy. Thus, although the conspiracy was vertical in nature, Apple as a vertical player was rightly held liable per se for violating ¤1 of the Sherman Act as a vertical player facilitat ing an already existing horizontal price fixing conspiracy. 14 Even if this case were subject to a rule of reason analysis, 14 See Toys" R" Us, Inc. v. FTC, 221 F.3d 928 (7th Cir. 2000) at ¦ 936. Price Output Q w Q a P w P a 2 A 1 B C P a & Q a : Retail price and output under agency P w Q w : Retail price and output under wholesale Decrease in Consumer Surplus Demand
Apple is still in violation of the Sherman Act because the results of the conspiracy are so plainly anticompetitive. On March 28th, consumers harmed by this conspiracy were granted class certification and are seeking $840 million in damages after trebling. More important than damages is the future of MFN clauses and agency contracts. This decision represents a recent trend of increased antitrust scrutiny being applied to MFN clauses. 15 16 Not all MFN clauses or agency contracts are starkly anticompetitive. Apple lawfully uses MFN clauses in many of its app negotiations with 3 rd party developers. Apple was guilty of a ¤1 violation because the MFN clause was used to eliminate retail price competition and raise the price of e books at the expense of consumers. Given the increased anti t rust scrutiny being applied, it will become increasingly important for businesses to weigh the procompetitive benefits of MFN clauses against any anticompetitive effects in order to achieve a balance and avoid antitrust violations moving forward. REFERENCES 1. PX 1097 Expert Report of Orley Ashenfelter in Connection with United States v. Apple Inc. et al., June 6t h, 2013 ("Ashenfelter Report"), retrieved from http://www.justice.gov/atr/cases/apple/exhibits/px 1097.pdf 2. PX 1105 Expert Report of Richard J. Gilbert in Connection with United S tates v. Apple Inc. et al., April 25th, 2013 ("Gilbert Report"), retrieved from http://www.justice.gov/atr/cases/apple/exhibits/px 1105.pdf 3. United States v. Apple Inc. et al, 952 F. Supp. 2d 638 (S.D.N.Y. 2013), retrieved from http://www.justice.gov/atr/cases/f299200/299275.pdf 4. Toys R" Us, Inc. v. FTC, 221 F.3d 928 (7th Cir. 2000). 6. United States v. Blue Cross Blue Shield of Michigan, 809 F. Supp. 2d 665 (E.D. Mich. 2011). 7. Starr v. Sony BMG Music Entertainment, 592 F. 3d 314 (2d Cir. 2010). 8. Susannah P. Torpey and Corina I. Bogaciu, "Practical Considerations in the Wake of the Apple Decision: How to Handle MFNs Going Forward," Antitrust Trade and Reg. Rep. (BNA) No. 105, at 196 (August 9, 2013) 9. Expert Report of Roger G. Noll in Connection with The State of Texas, et. al., v. Penguin Group (USA), Inc., et. al., October 21st, 2013 (" Noll Report"), retrieved from http://media.publishersmarketplace.com/wp content/uploads/2013/10/expert.pdf 15 See US v. Blue Cross Blue Shield of Michigan, 809 F. Supp. and Starr v. Sony BMG Music Entertainment, 592 F. 16 Susannah P. Torpey an d Corina I. Bogaciu, "Practical Considerations in the Wake of the Apple Decision: How to Handle MFNs Going Forward ," Antitrust Trade and Reg. Rep. (BNA) No. 105, at 196 (August 9, 2013)