DEPARTMENT OF ECONOMICS WORKING PAPER. Department of Economics Tufts University Medford, MA (617)

Transcription

1 DEPRTMENT OF ECONOMICS WORKING PPER 006 Deartment of Economics Tufts University Medford, M htt://ase.tufts.edu/econ

2 bstract dvertising: The Good, the Bad and the Ugly Lynne Peall, Dan Richards, and Liang Tan Deartment of Economics Tufts University Medford, M 0 lynne.eall@tufts.edu dan.richards@tufts.edu liang.tan@tufts.edu We model the choice of firms cometing in rices in a differentiated roducts maret to bundle advertising messages with their goods in return for ayment from advertisers. From the firms ersective, the otential to earn revenue from advertisers, maes advertising a good. However, because consumers in the roduct maret dislie such advertising, the bundling damens demand and in this sense is a bad. There is also a third role layed by advertising, however. Since a firm that bundles advertisements with its good sells a less attractive good, it has to rice more aggressively than one that does not do such bundling. Thus, bundling advertisements with the good can lead to more aggressive roduct ricing and thereby intensify roduct maret cometition. In this sense, advertising can mae things ugly.

3 dvertising: The Good, the Bad, and the Ugly By Lynne Peall, Dan Richards, and Liang Tan. Introduction dvertising is often, if not quite always, a bundled roduct. Until recently, such bundling has been done by media firms, notably, newsaers, radio, and television that combine entertainment roducts with advertising messages. Today, however, advertising is being bundled into an increasingly diverse range of consumer roducts, such as video games, movie screenings, cell hone services, and wraed cars. Recent examles include Microsoft who, in Setember 00, started sending mobile ads, showing the comany s logo when customers viewed certain Web ages from their handsets. In that same month, Visa US also launched a camaign in which consumers sending a five-letter text code to receive weather reorts get them with an ad-romoting Visa. Even more recently, Verizon Wireless and Srint Nextel have announced lans to test how consumers react to short video ads on their hones. What maes those traditional roduct-driven firms changes their strategies and begin to bundle ads into their roducts acting as firms in the media industry? In this aer, we rovide a simle duooly model that hels to isolate the factors behind this decision. The model investigates both the roduct maret and the advertising maret faced by the firms. When a firm decides to bundle another firm s ads into its roduct, it enters into a new maret the advertising maret. This could of course bring extra revenue to the firm. In this sense, advertising is a good. However, bundled ads are generally dislied by consumers. The ads interrut the consumtion of the roduct and thereby diminish consumer enjoyment. That is, the quality of the roduct is reduced when it is used as an advertising medium and this decreases the firm s roduct demand. In this sense, advertising is See Matt Richtel Their ds Here-mareters want to aear on the small screen. The New Yor Times, January 6, 006,. C

4 a bad. Note, however, that as Kind 00 et al, oint out, the rival of the firm that bundles advertising with its roduct may benefits from this activity because the relative quality of the rival s roduct is now enhanced. The decision to bundle advertising must therefore include a consideration of how this negative effect in the roduct maret will be enhanced by additional revenues in the advertising maret. We show how this asect of advertising can mae cometition more intense or, as the firms may view it, uglier. Previous authors, such as Liebowitz 98, Chae and Flores 998, Sumner 00, and Dues and Gal-Or 00 have also examined the bundling of advertising with news or electronic radio and television media. Somewhat closer to our wor is Godes, Ofe and Sarvary 00. They investigate how firms incentives to use their roducts as advertising media are affected by cometition in both the roduct maret and advertising maret. Their aroach though differs from ours in a number of ey ways. First, their basic model has firms cometing in quantities in an homogeneous roduct maret, and bundling advertising into the roduct does not adversely affect consumer valuation of the roduct. In equilibrium, all firms will always choose to bundle some ositive amount of advertising into their roducts. Moreover this outcome continues to hold even when they allow for consumer disutility from advertising. This outcome arises because the decision to advertise is not a strategic decision that affects how roducts are differentiated in the roduct maret. Rather a firm s decision to bundle advertising is equivalent to incurring an additional cost in the roduct maret. In equilibrium all firms choose to advertise and the amount of bundled advertising falls as the cost incurred increases. In our model, on the other hand, the roduct maret is a differentiated duooly where firms comete in rices, and a firm s decision to advertise affects not only consumer valuation of its roduct

5 but it also directly affects consumer valuation of the rival roducts. Secifically, if one firm decides to bundle advertising into its roduct then consumer valuation of its roduct falls but consumer valuation of a rival roduct increases. This strategic interaction on how advertising affects consumer valuation of all roducts in the maret is imortant, and suggests the ossibility of an asymmetric outcome where some firms bundle advertising and others do not. Thus, we include in the first stage of the game the firm decision of whether or not to advertise. This allows us to consider asymmetric outcomes in both the roduct and advertising marets. Kind, Nilssen and Soregard 00 also use a duooly model to investigate how roduct maret cometition affects a firm s incentive to use its roduct as a medium for advertising. They also have the two firms cometing in rices in the roduct maret and comete in rices in the advertising maret. However their aroach to modeling differentiation and the degree of cometition in the roduct maret is notably different from ours. In Kind, Nilssen, and Soregard 00 the degree of cometition is measured by whether the rival s roduct is a close or oor substitute. The less differentiated the roducts the tougher is rice cometition for the same grou of consumers. marginal cost. When the goods are erfect substitutes Bertrand rice cometition drives rice to In their model, it is the toughness of cometition in the roduct maret that underlies the firms incentives to use their roducts as advertising media. In contrast, our aroach to modeling rice cometition in the differentiated roduct maret ermits maret niches. In our model, when a firm s good is highly differentiated it faces a relatively small demand for its roduct. The firm has a maret niche, and as such its demand is not much affected by the rival firm s rice. This maes differentiation a double-edged sword. It softens rice cometition but limits the size of the maret. When a firm marets a less differentiated roduct, one

6 that is similar to the rival s roduct, firms comete intensely in rice for customers but there are more customers. result is that rice cometition in our model is not necessarily tougher with undifferentiated roducts because the greater rice sensitivity that increased roduct substitutability brings is offset by the fact that such greater substitutability is associated with a larger overall maret for the firms roducts. Since audience size or roduct maret sales is a ey factor for advertisers this feature of our model ermits us to cature an imortant relationshi that influences whether and how much advertising firms will bundle with their roducts. In sum, the revious literature on bundled advertising focuses mainly on media firms that are by assumtion active in both the roduct and the advertising maret. This literature largely sees to answer to the how much questions, i.e. the amount of revenues derived from the roduct and advertising marets, resectively. In contrast, our aer loos at the traditional roduct-maing firms and instead tries to answer the whether or not question, i.e., under what conditions does a consumer roduct firm, e.g., Verizon Wireless, start to bundle advertising into its roduct. s we shall see, our analysis uncovers critical risoner s dilemma roblems faced by these firms. This in turn has some imortant olicy imlications. The rest of the aer roceeds as follows. In the next section, we resent the formal model. Section discusses the model s imlications for both conventional, single-roduct firms and for olicy maers. Some concluding remars follow in Section.. The model The model is a two-firm and two-maret dynamic game. Firms comete in rices to sell their differentiated roducts in the roduct maret. Here, each has the same constant marginal cost of roduction, which, for simlicity, is set equal to zero. Firms also consider whether to use their

7 roducts as advertising media and earn additional revenue from the advertising maret. The roduct maret demand function is: q V, {0,}, i, ; j i i i j i j i where 0 < <, 0 < <, and V and is normalized to tae the value of one. Firm i s roduct demand deends on its own rice i as well as uon the rice of the rival j. The arameter measures the ositive effect on the firm s demand of a rice increase by the rival firm. It also measures indirectly how substitutable the two goods are. When is equal to zero, the two goods are comletely indeendent. s increases the two goods have a larger degree of substitution. Observe as well that the larger is, the larger is the firm s overall maret. Increased substitutability or less differentiation is associated with oerating in a bigger maret whereas low substitutability or a small value of is associated with selling a highly differentiated roduct in a relatively small maret niche. Each firm s demand is also affected by its advertising strategy, denoted by advertising strategy j. i or i and its rival s j taes the value of one if the firm chooses to advertise and 0 otherwise. The arameter measures the imact of the other firm s advertising choice on the firm s demand function. Because consumers generally dislie bundled advertising, the decision to include such bundled ads reduces the firm s roduct quality and therefore has a negative imact on its demand. On the other hand, if its rival does the bad advertising, it enhances the firm s relative roduct quality and thereby raises demand. Because the ads are bundled into the firm s roduct, the quantity of ads that the firm sells in the advertising maret is directly roortional to the quantity of the goods that the firm sells in the 6

8 roduct maret. For convenience, we normalize this factor of roortionality to one. The inverse demand function for advertising faced by the firms is: r V q q where. The rice r denotes the rice of advertising that firms receive for a bundle of advertising er unit roduct. The arameter measures the size of the advertising maret relative to the roduct maret. We assume that to ensure that the advertising maret is not too small to be attractive to the firms. If both firms advertise, the two firms comete in the advertising maret in a standard Cournot fashion. If only one firm advertises, it taes the whole maret. s before, firms have the same constant marginal cost of advertising, which we again set equal to zero. The game has two stages. In the first stage, each firm simultaneously decides whether or not to advertise, i.e., to bundle the bad in with its roduct. This is a discrete decision. In the second stage, the firms comete in rices, and, in the differentiated roduct maret having observed the advertising choices in stage one. We solve the game in the usual fashion by using bacward induction. Stage : In stage two, the advertising strategies, and have already been determined. Given demand for the differentiated roducti, q, ;,, i,, j i, firmi chooses rice to maximize its rofit π, ;, : i i i j i j i i j Maxπ q r q Maxπ q r q i j lternatively, we can use an extra arameter to cature the quantity of ads bundled into the roduct and let the firm decides this quantity. This however maes the model much more comlicated while adding little additional insight. 7

9 8 The first order conditions with resect to i for these maximization roblems imly the following best resonse functions: ] [ ] [ 6 Solving the functions and 6 together, yields Nash equilibrium rices, in stage. These are: 7 8 Corresonding to these rices are rofits: 9 π

10 π 0 Stage : In the first stage, the two firms simultaneously decide whether to bundle advertising into their roduct. This is a discrete yes or no decision. Either a firm advertises and sets i, or it does not and i 0. Observe that although the decision to bundle is discrete, the amount of advertising that firm i chooses when i is continuous, because one advertising unit is bundled with each unit sold in the roduct maret. In maing the choice of whether to use its roduct as an advertising medium, each firm anticiates the imlications of these choices in terms of the ayoffs in the stage two maret game in which each firm earns revenue from otentially two marets the roduct and the advertising maret. We may analyze the stage one game with the following strategic form matrix: Firm 0 i j Firm 0 π π 00 π 00 π π π π 0 π The ayoff π i associated with each air of strategies that the two firms choose is shown in the aroriate cell of the matrix. Using equations 7,8, 9 and 0it is easy to show that the equilibrium roduct rices and rofits in each ossible strategy combination are as follows:

11 π π B 0, 0 8 π π C , π 0 π 0 0 0, 0 8 π 0 π 0 8 D π π Using the iterated elimination of strictly dominated strategies and comaring the rofits under different air of strategies, we find that there are four ossible Nash equilibria based on following raning of the i j rofits π i : If π > π 00 and π 0 > π 0, then neither firm bundles advertising, or 0. If π > π 00 and π 0 < π 0, then two equilibria exist. In one neither firm bundles advertising, or 0, whereas in the second both firms do, or. If π < π 00 and π 0 > π 0, then there are two asymmetric equilibria. One firm bundles advertising whereas the other one does not, or, 0 in one and 0, in the other. If π < π 00 and π 0 < π 0, then both firms bundle advertising or. The main urose of the aer is to answer the question of what drives traditional 0

12 roduct-driven firms to act as advertising media. The exression of the equations - and the analysis of otential Nash equilibrium show that this choice deends on the arameters,, and. Unfortunately, this is a comlex arameter sace to wor with and one not easily suscetible to closed form solutions. To analyze the roblem and to see how changing the arameter values leads to different Nash equilibria, we examine the relationshi between and while holding constant at. We then increase by increments 0. and reeat the rocess. In each case, we set π π 00 and 0 π. In this way, we can, for the given value, lot and and derive the two π 0 critical lines π π 00 and π 0 π 0. The areas divided by these two lines reresent the regions in which the different Nash equilibria will obtain. We analyze how these regions change as increases. These results are illustrated by the various grahs in Figure.. Discussion of the results We organize our discussion of the results illustrated in Figure in the following way: First, we resent our main roosition and then briefly discuss the results of our findings regarding the relative size of the advertising maret as reflected in the arameter. We then describe briefly the conditions under which the firms confront a risoner s dilemma in stage one and the extent to which the failure to maximize roducer surlus is equivalent to a failure to maximize total surlus i.e. consumers are also adversely affected in the risoner s dilemma. This welfare analysis ermits us to draw some tentative olicy conclusions. We also comment briefly on the imlications of the model for roduct rices.. The size of the advertising maret relative to the size of the roduct maret s noted, the arameter is a rough way to catures the size of the advertising maret relative to the size of the roduct maret. s the various grahs in Figure show, this arameter is an imortant determinant of whether firms have an incentive to enter the media maret and bundle advertising into

13 their roducts. s increases, it becomes more and more liely that both firms will enter the maret and engage in such bundling. For relatively low values of., the only equilibrium is one in which neither firm enters and bundles advertising into its roduct. s gets larger, it is ossible that at least one firm enters the media maret and includes bundled advertising. This initially occurs in the uer left corner of the, arameter sace where is small and is large. In this region the ositive effect on the rival s demand of bundling is relatively large whereas the effect of the rival s rice is relatively small. So if one firm enters the advertising maret and the rival firm does not then the quality of rival firm s roduct increases but the two roducts are not very substitutable. s increases and the otential revenues from advertising rise, it becomes much more liely that the equilibrium will include at least one firm bundles advertising with its roduct. For still larger values of, the arameter sace includes a region in which both firms choose to enter the advertising maret. This initially occurs in the region in which both and are small. In this region the one firms decision to advertise does not sillover much to the other and the roducts are not that closely substitutable. s increases this region exand. When is.6, it covers 80 ercent of the, arameter sace. When, it covers 00 ercent, and the outcome is that in equilibrium both firms enter the advertising maret and bundle advertising into their roducts. We summarize our results with the following roosition the roof of which is given in the endix. Proosition a When the advertising maret is relatively small, <.690, no firm will enter the advertising maret and bundle advertising with its roduct. When the advertising maret is relatively large,, both firms enter the advertising maret and will bundle advertising. For values of such that.690 < <, various equilibria are ossible in which either no firm, one firm or both firms advertise.

14 . Prisoner s dilemmas, roducer surlus, and social welfare s the grahs in Figure show there are values of the arameters,, and, for which the Nash equilibrium will be one in which neither firm includes bundled advertising with its roducts or both do. Some of these equilibria, from the oint of view of firm rofitability, are not first best outcomes. Instead, they reflect risoner s dilemma tye outcomes and as such both firms could be made better off if they could cooerate and achieve another outcome. To determine whether the game is a risoner s dilemma, we comare π and π 00 in each game. gain, in the - lane, we draw the critical curve along which π π 00. This curve searates the 0 equilibrium region into risoner s dilemma and non-risoner s dilemma categories. First, we investigate the non-advertising case that arises for relatively small.. Figure a illustrates this outcome for the case in which is exactly.. s shown, the uer left art of the arameter sace shaded in green is the region of the risoner s dilemma. lthough neither firm has entered the advertising maret, and neither bundles advertising in its roduct, both would be better of if each did. The underlying intuition is straightforward. When the two goods are less substitutable and the consumer base is not that large but the ositive sillover from the rival entering into advertising is relatively large, then if one firm enters into advertising the other firm finds more rofitable not to enter. The outcome is that neither firm advertises but if both did they would both be more rofitable. The no-advertising equilibrium may also not be best for society overall. Figure b again shows the case of a relatively small advertising maret,.. s before, everything to the northwest of the lower curve, here shown as the combination of a green-shaded and orange-shaded

15 region, indicates the area in which roducer surlus is lower than it would be if both firms advertised. The orange-shaded subset of this region is the ortion of this sace in which not only roducer surlus but the total surlus is reduced because neither firm advertises. This region reveals a otential for maret failure in which there is no advertising even though both roducers and consumers could otentially be better off if both firms bundled advertising with their roducts. Recall consumers in the roduct maret can benefit from the lower rices, or erhas free or even subsidized consumtion of the roduct when both firms enter into the advertising maret. s increases and equilibria occur in which both firms include bundled advertising, the risoner s dilemma cases and otential social welfare losses occur when both firms enter the advertising maret. Figures a and b illustrate this ossibility for the case of.9. Here, the green-shaded region of Figure a indicates that ortion of the arabola-shaed region of the arameter sace in which both firms advertise but where roducer surlus would be larger if neither did. The large, orange-shaded ortion in Figure b then shows the ortion of the shaded region in a for which this failure to maximize total roducer surlus also means that the total surlus is not maximized. s the figure suggests, the region in which total welfare would be raised by a ban on bundled advertising is relatively large, esecially in comarison with the region in which such a ban would raise roducer surlus. This is a common feature of all the cases in which bundled advertising is excessive. It suggests that ublic olicy to ban bundled advertising will very frequently be in the ublic interest. Figures a and b illustrates the otential for excessive advertising for those cases in which the only ossible Nash equilibria are those in which both firms enter the advertising maret and choose to bundle advertising into their rimary roducts. s in Figures a and b, there are many

16 arameter configurations in which firms would do better if there were some statutory ban that rohibited them from including advertising with their roducts. Here again, just as in the earlier figures, there is a large lielihood that such a ban would raise welfare overall.. Pricing In discussing the various cases, we have not exlicitly addressed the nature of the equilibrium roduct rices. In any equilibrium in which neither firm advertises, it is clear that roducts are goods and rices are ositive. However, in an equilibrium in which at least one firm advertises, there is the ossibility that the deressing effect this has on consumer roduct valuation could lead to the roduct being a bad and a negative roduct rice. Negative ricing should not in this case be ruled out. Effectively negative ricing amounts to aying the consumer to listen to the bundled ad, and this henomenon is frequently observed. For examle, real estate firms often ay consumers by giving them goods in return for listening to an hour-long resentation on a time-sharing oortunity. Similarly, so-called wraed cars give consumers the free use of an automobile but one that is covered in advertising. The Disney Comany offers yet another examle. In early 00, it gave consumers free access to a number of on-line games that were based on Disneyland attractions e.g., the Haunted Mansion in an effort to lure consumers to the theme ar in celebration of Disneyland s 0 th anniversary. ccordingly, we have not rule out equilibria that include negative roduct rices. Moreover as gets large and entering the advertising maret and selling a lot of bundled advertising becomes increasingly rofitable, the ortion of the, arameter sace that includes such negative ricing also increases. If we restrict attention to only those equilibria with non-negative roduct rices then the roduct becomes accordingly a free good in a great many cases. See Michael McCarthy Disney lans to mix ads, video games to target ids, teens US Today, January 7, 00.

17 . Conclusion Increasingly the age of the information highway is one in which information is bundled with other goods. In recent years we have witnessed more and more traditional roduct firms maing a decision to bundle advertising with their roducts and thereby imlicitly serve as a communications or information medium. In this aer, we use a simle duooly model to study this henomenon. We assume the two firms comete in rices in the roduct maret and comete in quantity in the advertising maret. By bundling or selling advertising sace in their roducts, firms gain additional revenues. However, such bundling reduces consumers valuation of the firm s roduct leading consumers to switch to the firm s cometitor while the bundling firm s roduct demand decreases accordingly. We show that the equilibrium deends on three ey arameters measuring, resectively, the relative size of the advertising maret, the degree of substitution between the rival roducts or the size of the consumer base, and the imact that bundled advertising has on roduct maret demand. When the advertising maret is relatively small, not surrisingly firms do not have an incentive to enter the advertising maret. This is true even when both firms would be more rofitable, and total surlus would rise, if each firm entered the advertising media maret and did engage in such bundling. s the size of the advertising maret becomes relatively larger, the lielihood that at least one firm will enter the advertising maret and use its roduct as an advertising medium increases. threshold value of the relative size of the advertising maret is reached beyond which the only ossible equilibria are those in which both firms enter the advertising media maret and bundle advertising with their roducts. Here as well there is the ossibility of a maret failure in which there is too much advertising and firms would both be better if neither entered the advertising maret. Moreover, 6

18 the arameter range over which there is a consumer welfare loss because of too much advertising is relatively large. n interesting feature of many equilibria in which both firms advertise is that roduct rices are negative, or constrained to be zero. The roduct firms have an incentive to ay eole or to give their roducts away to use with a view toward increasing circulation and raising advertising revenue. This seems to be consistent with a number of real world cases involving such give-a-ways, such as those emloyed by real estate firms. 7

19 . References Chae, Suchan, and Daniel Flores. "Broadcasting Versus Narrowcasting." Information Economics and Policy 0, no. March, 998: -7. Chaudhri, Vive. "Pricing and Efficiency of a Circulation Industry: The Case of Newsaers." Information Economics and Policy, 0, no. March, 998: Dues, nthony, and Esther Gal-Or. "Negotiations and Exclusivity Contracts for dvertising." Mareting Science, no. Sring 00, 00: -, Godes, D., E. Ofe, and M. Sarvary. "Products Vs. dvertising: Media Cometition and the Relative Source of Firm Profits." Harvard Business School Mareting Research Paer Kind, H. J., T. Nilssen, and L. Sorgard. "Financing of Media Firms: Does Cometition Matter?" Woring Paer 00. Liebowitz, S. J. "The Imacts of Cable Retransmission on Television Broadcasters." Canadian Journal of Economics, no. ugust 98, 98: 0-. Sumner, D. E. "Who Pays for Magazines? dvertisers Or Consumers?" Journal of dvertising Research 00:

20 6. endix Proof of roosition : For neither firm to include bundled advertising requires that rofits when neither advertises exceed those when at least one advertises, i.e., that π 00 π. We first solve for the conditions 0 under which these two values are equal. This imlies: 8 Setting, yields: or is not a valid solution, as is negative when 0< <. Minimize in by setting 0 yields: The numerical solution is.690, when 0< <. This is the minimum value of for which π π in the ermissible 0, sace of and. Therefore, when <.690, no 0 00 advertising is the only ossible equilibrium in the maret. 9

21 n equilibrium in which both firms advertise requires that the rofit either gets in this state exceeds the rofit one would earn if that firm unilaterally stoed bundling advertising with its 00 roduct. That is, we must have π π Now return to equation and set 0 0, to obtain: or The solution in is not valid as is negative when 0< <. Focusing on the solution in 6 and setting, we obtain.. This is the maximum value for which the critical line 00 π 0 π is still in the 0, sace of and. ny larger than that will mae the asymmetric equilibrium not ossible in the maret. long this critical line we have π π, or Set 0 yields or

22 Only 8 is a valid solution consistent with 0< <. Now, minimize by setting 0. The numerical solution is then.77, when 0< <. This is the minimum value of for which the critical line π π 0 lies in the ermissible 0, arameter sace of and, i.e., for which both firms advertising, as oosed to just one, becomes a ossible equilibrium. In a similar fashion, it is easy to show that the oint at which this line lies to the southeast of the entire ermissible arameter sace is when.

23 .: Neither dvertises Figure.: Either One dvertises Or Neither dvertises No dvertising advertises No one advertises : Either One dvertises Or No One dvertises.9: Either Both dvertise, One dvertise, or No One dvertises One advertises No one 0. One advertises advertises Both advertise No dvertising : Either Both dvertise Or.0: Both Firms Just One dvertises dvertise Both dvertise One advertises Both dvertise

24 Figure a Small size advertising maret,. Shaded area reflects region of Prisoner s Dilemma Neither dvertises but Both Should Figure b Potential for Maret Failure Due to Insufficient dvertising When Figure a Medium size advertising maret,.9 Shaded area reflects region of Prisoner s Dilemma Both dvertise but Neither Should Figure b Potential for Maret Failure Due to Excessive dvertising When.9 Total Surlus rises in Orange-Shaded rea 0.8 One advertises 0.8 One advertises Both advertise No advertising Both advertise No advertising Figure a Large advertising maret,.0 Shaded area reflects region of Prisoner s Dilemma Both dvertise but Neither Should Figure a Potential for Maret Failure Due to Excessive dvertising when Both dvertise but in Green-shaded rea, neither should Both dvertise, Producer Surlus rises in Green Orange rea, Total Surlus rises in Orange rea

25 WORKING PPER SERIES 006 htt://ase.tufts.edu/econ/aers/aers.html IONNIDES, Yannis and driaan SOETEVENT; Wages and Emloyment in a Random Social Networ with rbitrary Degree Distribution DGEN, Richard and Daniel RICHRDS; Merger Theory and Evidence: The Baby-Food Case Reconsidered PEPLL, Lynne, Daniel RICHRDS, John STRUB and Michael DEBRTOLO; Cometition and Civic Engagement in the Religious Maretlace LOURY, Linda; Job Search mong Informal Contacts BIR, nua and Karen EGGLESTON; Measuring Selection Incentives in Managed Care: Evidence from the Massachusetts State Emloyee Insurance Program PEPLL, Lynne, Daniel RICHRDS and Liang TN; dvertising: The Good, the Bad and the Ugly METCLF, Gilbert E.; Tax Incidence METCLF, Gilbert E.; Value-dded Tax METCLF, Gilbert E.; Energy Conservation in the United States: Understanding its Role in Climate Policy LOURY, Linda; ll in the Extended Family: Grandarents, unts, and Uncles and Educational ttainment IONNIDES, Yannis; Emirics of Social Interactions.