A Critique of the FTC'S Analysis of the Proposed Tobacco Settlement by AAI Advisory Board Member Robert L. Steiner, followed by debate with Jeremy Bulow and Jonathan Baker

On November 22, 1999, Bert Foer wrote an FTC:WATCH column reporting on a new paper by AAI Advisory Board Member Robert Steiner, in which Steiner criticized two FTC Staff Reports on the tobacco industry. Steiner raised important methodological questions about the nature of pass-throughs when a manufacturer's costs are increased. The Foer column follows below and the Steiner paper immediately after. On January 10, 2000, FTC Bureau of Economics Director Jeremy Bulow and former Director Jonathan Baker published a reply in the FTC:WATCH. That appears below the Steiner paper. Next, Steiner replied in the FTC:WATCH of January 18,2000, and this appears at the end of the current string, below the other articles.


By Albert A. Foer

Robert L. Steiner has written a provocative critique of two FTC Staff Reports, Competition and the Financial Impact of the Proposed Tobacco Industry Settlement (September, 1997) and Evaluation of the Tobacco Industry Analysis Submitted to Congress on October 8,1997(November 10, 1997). Congress had asked the FTC to predict the economic effects of the proposed tobacco industry settlement. You will recall that the settlement, which ultimately fell apart and never obtained Congressional approval, would have required the tobacco companies to make annual payments to fund various federal and state programs that were expected to raise the factory price of cigarettes. The principal purpose of the Commission' s investigation was to estimate the magnitude of the resultant increase in the retail price per pack of cigarettes and of the consequent decline in cigarette consumption.

Before reporting on Steiner's analysis, it is useful to know something about his personal background. Trained in economics, Steiner elected to enter his family's consumer goods manufacturing businesses. At different times, these included soap, over-the-counter drugs, soft drink concentrates, and toys. After twenty-five years, during which he became president of the toy company Kenner Products, Steiner retired to accept a teaching position at the Graduate Business School, University of Cincinnati. He subsequently was invited to the FTC's Bureau of Economics, where he served for five years as a senior staff economist. Presently involved in consulting and writing, Steiner serves on the Advisory Board of the AAI.

Steiner has authored numerous articles exploring various aspects of the consumer goods economy. Three are referenced at the end of this note. What has motivated Steiner has been the disconnect he perceives between much of economic theory and his own business experiences and subsequent research. One of his insights has been the recognition that in consumer goods markets, "Both the first-stage manufacturer/retailer market and the second-stage retailer/consumer market are imperfectly competitive." Antitrust economists, for the most part, have ignored this, postulating instead a "single-stage" framework, "in which the market downstream from the consumer goods maker tacitly is assumed to be inert and perfectly competitive, unless it is cartelized or monopolized." In his various writings, Steiner has fashioned an informal "dual-stage" model in which to capture and analyze the horizontal relationships among firms at the two stages and the vertical relationships between firms at different stages. He has used this framework to explain and predict the economic effects and the welfare implications of manufacturers' advertising, of productivity changes, of vertical restraints, and other topics in consumer goods industries.

Steiner reports that the FTC's "baseline model" estimates that after 5 years cigarette manufacturers' prices would be $.57 higher than they would be without the proposed settlement. The FTC posited that the wholesaler would pass through the $.57 price increase to the retailer who would pass it through to the consumer. So the trade would pass through the increase in cigarette manufactures' price without adding any further markup. Right?

Wrong. According to Steiner: "Certainly the downstream markets in this industry are quite competitive. Yet one cannot wave a single-stage wand over them and declare them perfectly competitive. They are, in fact, monopolistically competitive markets in which firms face downward sloping demand curves. Most theories of monopolistic competition provide that a downstream firm would seek to take a mark up on the total of its invoice and non-invoice marginal costs such that the margin on its selling price is the inverse of the perceived elasticity of demand. Thus the investigator must estimate the elasticity of the representative wholesaler's and retailer's demand curves and their marginal or average variable costs. Together, these parameters determine the extent to which the new retail price will be marked up over the new factory price along the industry's consumer level market demand curve. The FTC's single-stage procedure erroneously omits this step and thereby underestimates the rise in retail prices and the fall in quantity sold."

Let me rephrase this, from my years of experience as a jewelry retailer. Few if any markets are more competitive. But, no matter how competitive the market, if our suppliers raised their price to us and all we did was pass that increase on to our customers dollar for dollar, we would soon be out of business, because our percentage profit margins would continually shrink and the dollar contribution margin would decline, the quantity sold going down as the price goes up. Various categories of merchandise had different gross margin targets because all categories are not equally competitive and therefore result in differing gross margins. Our buyers were evaluated on the basis of their ability to meet their category targets. Thus, they would apply the appropriate category markup to the new cost of goods. If, very occasionally, competition forced us to reduce a price, we would do that later on.

In every industry the competitive dynamic produces a certain markup for each category of goods. In talking to tobacco wholesalers and retailers, Steiner found that when cigarette makers had previously raised their prices, the trade had marked up the price increase sufficiently to nearly maintain pre-existing percentage gross margins. He then analyzed the effects of the $.45 per- pack increase that cigarette manufacturers' instituted in November, 1998, in connection with a settlement of the states' law suits. The FTC Report estimates the cigarette industry's 1997 trade margin (the sum of the dollar markups of wholesalers and retailers) at $.47, which Steiner found was higher than estimates he obtained from other sources. Based on his empirical data, his trade margin equation states that the new trade margin would be 1.256 times the pre-price-increase margin (-) $.02. Using this equation and the above FTC estimate, the distributive trade added another $.10 to the $.45 factory price increase, causing the retail price of a pack of cigarettes to rise by $.55. Steiner also found that the percentage gross trade margin had only declined trivially following the $.45 factory price increase.

Consequently, it appears that the FTC underestimated the price increase that would face consumers and also underestimated the likely reduction in consumption. If Steiner is correct, many economists will have to reconsider their methodology for predicting the impact of changes in upstream prices on the prices that consumers will pay. There is also an implication for the calculation of damages in antitrust cases involving manufacturer price-fixing: consumers who are deprived of standing by Illinois Brick are injured not only by the monopoly rent, but by the distributive trades' additional markup on that rent-exacerbating the injustice of Illinois Brick.

(Steiner's paper on the tobacco settlement is available on the AAI website, www.antitrustinstitute.org. It is part of an article entitled "The Third Relevant Market," forthcoming in the Antitrust Bulletin. For those interested in Steiner's contribution to the literature of the consumer goods economy, see: "Intrabrand Competition-Stepchild of Antitrust," 36 Antitrust Bulletin 155 (1991);"The Inverse Association Between the Margins of Manufacturers and Retailers," 8 Review of Industrial Organization 717 (1993); and "How Manufacturers Deal with the Price-Cutting Retailer: When Are Vertical Restraints Efficient?" 65 Antitrust Law Journal 407 (1997).)



By Robert L. Steiner

Over 30 years ago the economist R. B. Heflebower complained that " in the development of price and output theory, economists have dealt primarily with manufacturing industries... and have treated these as though manufacturers sold finished products directly to consumers, or alternatively, as if the distributive trades were analytically neutral. " ( Heflebower, Internal Trade in International Encyclopedia of the Social Sciences, David Sills, Ed. (1968) at 49 emphasis added). This is to say that wholesalers and retailers buy and resell as perfect competitors and face perfectly elastic, horizontal demand curves.

In the above "single-stage" world changes in factory prices are passed through on a dollar-for-dollar basis into retail prices. Then, armed with an estimate of the price elasticity of consumer demand and the new retail price, the fall in quantity demanded by consumers is handily calculated. In a series of articles I have attacked this approach as producing highly inaccurate and often biased estimates of outcomes in a multi-stage world where the foregoing conditions are not present, (Robert L. Steiner, The Inverse Association Between the Margins of Manufacturers and Retailers, Review of Industrial Orizanization 8: 717-740, 1993 and the earlier articles referenced there). Unfortunately, despite some undoubted advances, the legacy of the "analytically neutral" distribution system lives on, even among highly skilled microeconornists. It was manifest in the analysis that Congress asked the FTC to prepare in 1997 of the economic effects of the proposed tobacco industry settlement.

The settlement required the tobacco companies to make annual payments to fund various federal and state programs that were expected to raise the factory price of cigarettes. The focus of the Commission' s investigation was to estimate the magnitude of the resultant increase in the retail price per pack of cigarettes and of the consequent decline in cigarette consumption. FTC Staff produced two Reports, Competition and the Financial Impact of the Proposed Tobacco Industry Settlement (September, 1997 ) and "Evaluation of the Tobacco Industry Analysis Submitted to Congress on October 8,199 7(November 10, 1997). In a theme that is reiterated in the November Report (at 6,note 16), the September Report (at 27 note 72) sates that the proposed settlement would not change the highly competitive nature of cigarette distribution- e.g. the elasticity of wholesalers' and retailers' demand schedules - nor the non-invoice costs of these downstream firms. Note 72 acknowledges that studies have shown that after increases in state excise taxes, which raise wholesalers' invoice prices to retailers, " retail prices may rise somewhat more than the tax." Yet, it concludes that " Because the wholesale and retail distribution sectors are competitive, any revenue increases that accrue after a tax increase must benefit the manufacturing sector" which is far from competitive. Hence, the $ trade margin will remain constant.

The FTC's "baseline" model does not explicitly identify the magnitude of trade margins. But it predicts that after 5 years the proposed settlement would cause the factory price of a pack of cigarettes to be $.57 higher than it would be without the settlement, causing the retail price also to increase by $.5 7 to $2.62 (November Report at 5, note 12). Applying the (-) 0. 4 elasticity of consumer demand in the FTC model, results in a sizeable, I I.1%, drop in quantity sold. Although it is difficult to identify the assumptions about factory prices in the Staff Reports, I read the data to imply that the $.57 hike would involve an increase of around 45% in selling prices to the trade. Under these circumstances, it seemed inconceivable to this former consumer goods manufacturer that wholesalers and retailers would not boost their dollar margins by a substantial amount. To do otherwise would be to suffer an important drop in their dollar contribution margins, even granting the validity of the FTC's assumptions of no increase in their non-invoice marginal costs (such as the cost of financing a unit of inventory and the "shrinkage" from theft) and no decrease in the elasticity of their demand schedules - assumptions that the trade finds contra factual (see below).

Certainly the downstream markets in this industry are quite competitive. Yet one cannot wave a single-stage wand over them and declare them perfectly competitive. They are, in fact, monopolistically competitive markets in which firms face downward sloping demand curves. Most theories of monopolistic competition provide that a downstream firm would seek to take a mark up on the total of its invoice and non-invoice marginal costs such that the margin on its selling price is the inverse of the perceived elasticity of demand. Thus the investigator must estimate the elasticity of the representative wholesaler's and retailer's demand curves and their marginal or average variable costs. Together, these parameters determine the extent to which the new retail price will be marked up over the new factory price along the industry's consumer level market demand curve. The FTC's single-stage procedure erroneously omits this step and thereby underestimates the rise in retail prices and the fall in quantity sold.

A helpful first step is to examine the numerous Commerce Department and private data estimates of wholesale and retail percentage gross margins. Properly weighted, the sum of these two margins is the % gross trade margin. It is a biased-high estimate of the true trade margin because it does not include the trade's non-invoice variable costs. Moreover, in long run equilibrium the gross trade margin must be high enough to cover downstream firms' fixed capital investments as well as their variable costs - otherwise they could not remain in business. Hence, when only marginal costs increase, we would expect the new equilibrium % gross trade margin to fall somewhat, although $ trade margins would rise.

Space limitations prevent more than a sparse summary of cigarette trade margins here. Based on various 1992 Censuses, cigarettes makers distribute about 3 / 4 of their output through wholesalers. Tobacco and Tobacco Products Merchant Wholesalers account for 86.3% of the sales of that commodity and had gross margins of 10.4 % on selling price. Federal excise taxes included in manufacturers' prices are part of purchases, but the state excise taxes paid by wholesalers and included in the price they charge retailers are not counted as sales. Retailer margins are harder to pinpoint, since so many kinds of retailers sell cigarettes and there are a plethora of coupons and off-price deals by cigarette makers. A group of aggressive outlets use cigarettes as traffic-builders and sell them at gross margins of 10% and less, Gross margins in tobacco stands/stores (Census KB # 599300) have averaged around 30%.Those of traditional supermarkets, perhaps the largest single outlet for cigarettes, have risen to the high teens.

The proposed 1997 tobacco industry settlement was eventually rejected by Congress, but the concepts raised above are important to the way the Commission models consumer goods industries in the future. (For a more in depth treatment of this issue see Steiner, The Third Relevant Market, Antitrust Bulletin, Forthcoming in 2000.) Selecting between the FTC's methodology and the one I advocate is a theoretical problem that must be informed by empirical evidence. A highly relevant test of the two theories' predictions was soon to come. In 1998 the tobacco companies reached an agreement with the states, the cost of which lead the major producers to raise the factory list price of a pack of cigarettes by $.45 in late November. Based on the history of factory price hikes, Darryl Jayson, long-time vice-president of the Tobacco Merchants Association (TMA), predicted that downstream firms would markup the increase by 20%-30%.When on August 30 cigarette makers announced a further $.18 increase, the consensus estimate of tobacco industry analysts in leading investment banking firms was that retail prices would increase by $.22, a 22.2% markup on the factory price hike. (Justin Pope, Cigarette Packs to Jump 22 Cents, Associated Press, reported in Yahoo Finance, Aug. 31,1999). Without exception, the tobacco wholesalers and retailers to whom I spoke felt that the FTC was misinformed, to put it politely, in its conclusion that they and their competitors had previously or would now pass through sizeable increases in their invoice costs on a dollar-for-dollar basis.

A senior executive of a substantial size supermarket chain, upon condition of anonymity, provided the following interesting information: I- His chain had marked up the $.45 increase to approximately maintain its % gross margin. 2- State "Unfair Sales Laws", not mentioned in the FTC Staff Reports, play a major price-increasing role. (TMA advises that 27 states have these laws which specify minimum margins for wholesalers and retailers and minimum retail prices). The off-price outlets that sell at the minimum retail prices must raise their prices roughly proportionately when factory prices rise. This permits competing higher margin stores, such as those of this supermarket, also to raise their prices by more than the increase in their invoice cost. 3- The executive took issue with the FTC's assumption that the elasticity of retailers' demand curves would not change. At the sharply higher prices per pack, individuals with more elastic demands would quit smoking and the "hooked" smokers with less elastic demands would dominate the reduced consumer pool. That will lead to both higher everyday retail markups and prices and to fewer cut- price, traffic- building newspaper adds.. 4- The executive disputed the FTC's view that the trade's non-invoice costs are unaffected by towering factory price increases. With higher invoice costs and slower turnover, interest costs per unit of inventory increase. Theft is a huge problem in cigarettes that is exacerbated as retail prices rise. Most theft is by individuals who steal to resell cigarettes, not by employees or shoppers for their own consumption.

Was the $.45 per pack factory price increase in late November, 1998 passed through to consumers on a dollar-for-dollar basis, per the FTC model, or by the time of the next increase in late August, 1999 had trade margins and retail prices risen by far more than that? Estimates of national average trade margins and retail prices- per- pack vary enormously because of the myriad manufacturers' couponing and pricing deals and the large inter- state variances in excise taxes and in minimum retail prices under their Unfair Sales Laws. Even the Bureau of Labor Statistics does not publish a retail price-per pack, but we can use its Consumer Price Index for cigarettes for November, 1998 (which does not include the $,45 price increase) and for August, 1999 (which does not reflect the subsequent $.18 increase.).

During this period the CPI, net of retail sales tax, rose 25.6%, the federal excise tax remained at $.24, and per TMA the national average state excise taxes rose from $.39 to $.43 per pack. I am indebted to TMA for its careful calculation of the average producers' price- per-pack. This was derived by multiplying the printed manufacturers' list prices in the premium, discount and deep discount segments by the market share of each segment during these two months. The prices were then decreased by the estimated 2% average cash discount given the trade on the sum of the published price and the federal excise tax. The cigarette manufacturer's invoice price to the trade is the sum of the producer price ($1.181 in November, 1998 and $1.625 in August, 1999) and the $.24 federal excise tax.

Our methodology does not per se produce a $ trade margin (TM) or a retail price- per- pack (P) but enables us to determine how the $ trade margin and the gross trade margin (TM / P) changed in response to the $.45 increase in manufacturers' prices. We designate November, 1998 by the subscript n and August, 1999 by the subscript a in the 2 equations below.

Nov. 1998 $1.181 +.24 +.39 + TMn = Pn.

Aug. 1999 $1.625 +.24 +.43 + TMa = Pa = 1.256 Pn

We multiply the top equation by 1.256 and set the two equations equal to each other to solve.

Result I TMa = $1.256 TMn (-) $. 02.

We can thus confidently reject the FTC's prediction of a dollar-for-dollar pass through, which requires a constant $ trade margin. Jeremy Bulow and Paul Klemperer calculate that the trade margin for 1997 assumed in the FTC Reports was $.47.(Table 6, The Tobacco Deal, Brookings Papers in Microeconomics 1998). Estimates from other sources are lower. If we apply the equation in Result 1 to the FTC's estimate, the trade added another $.10 per pack to the $.45 factory price increase - causing retail cigarette prices to rise by $.55.

To determine what happened to the gross trade margin (TM / P), we divide the left side of Result I by Pa, the first term on the right side by $1.256 Pn and the second term by Pa.

Result 2 GMa = GMn (-) $.02 / Pa.

With an average August, 1999 retail price-per- pack in the range of $2.50 to $2.95,the gross trade margin dipped only very slightly from its pre-price-increase level, instead of falling rather substantially as predicted in the FTC model. Both Result I and Result 2 are inconsistent with the single-stage paradigm and have important implications for the way consumer goods industries should be analyzed.


December 20, 1999


To the Editor:

We welcome Albert A. Foer and Robert L. Steiner's close reading of the two FTC staff reports on the originally-proposed tobacco settlement because we inviteconstructive debate. [FTC: WATCH No. 533, November 2].But we take issue with much of their critical analysis and think the FTC staff has made by far the better case.

The dispute is about whether an industry-wide cost increase to the distribution sector of the cigarette industry - wholesaling and retailing, but not manufacturing - would be passed through to consumers with or without an additional markup. If cigarette manufacturers raise their wholesale price by 45 cents per pack, for example, will retail prices rise by about 45 cents (the FTC staff report's view) or by substantially more than 45 cents (the tobacco industry's view, and Foer and Steiner's position)? The FTC staff report explained that an industry-wide cost increase will be completely passed-through to consumers, without an additional markup, if the industry is competitive and firms have constant marginal costs. This is, after all, how the distribution sector of the cigarette industry looks: in most localities, cigarette consumers can choose among a large number of convenient alternative sellers when purchasing that product, and there is no reason to think that incremental costs would vary for most wholesalers and retailers were product sales to decline.

Foer and Steiner challenge the FTC staff's view on both theoretical and empirical grounds. Their theoretical argument is demonstrably incorrect. Even if the distribution sector should be thought of as monopolistically competitive rather than perfectly competitive, it is simply not true that "most theories of monopolistic competition" predict that higher industry-wide input costs (such as an increase in the excise tax) will generate a price increase even greater than the cost increase. So long as demand is such that a parallel upward shift in the demand curve will lead to an increase in market prices - a condition commonly satisfied - then the pass-through rate will be less than one hundred percent in a monopolistically competitive market. Thus, the "common sense" view that if a retailer's costs go up, it will necessarily raise price to maintain the same percentage price-cost margin is simply wrong. Moreover, if Foer and Steiner's estimates are correct, then a recent wholesale price rise is increasing retailer margins by over 20 percent while most estimates are that demand will fall by about 10 percent. So the retailers would be making more money because their costs rise! If so, then retailers of cigarettes and other consumer products with inelastic demand should be imploring legislatures to increase their costs by raising excise taxes in order to boost their profits.

Foer and Steiner commendably refrain from resting their argument on theory alone, but seek out empirical evidence on actual pass-through rates in the cigarette industry. Aside from anonymous retailer anecdotes, however, their empirical evidence consists entirely of a single suspect observation. They claim to have inferred from changes in the consumer price index for cigarettes that the distribution sector responded to a 45 cent per pack increase in the wholesale price in late 1998 by raising retail prices over the next nine months by 55 cents per pack. If this inference is correct, the distribution sector passed through 122% of the cost increase, or 10 cents per pack more than the 100% pass through rate the =46TC staff would have predicted. What could be wrong with this conclusion?

Although we have not been able to evaluate their work in detail, we have a number of concerns. Foer and Steiner evaluate the November 1998 wholesale price increase by looking at CPI data for August 1999. But using slightly different time periods, for example October 1998 to October 1999, would have yielded dramatically lower pass-through rates. Also, looking at one observation over a nine month period does not allow for the possibility that other factors may have affected margins. Furthermore, the posted wholesale price may not be a good measure of actual dealer cost, and the CPI may not be a good measure of price changes because it does not account for changes in coupon promotions and may incorporate biases resulting from the way it accounts for differences in brand mix across cities.

Neither Foer and Steiner's theoretical analysis nor their empirical evidence provides much reason to question the FTC staff's conclusion that the cigarette distribution sector would pass through increases in the wholesale price without tacking on an additional price increase. Without more, we are unwilling to take on their challenge to reconsider the economic methodology for predicting the rate at which changes in the wholesale price are passed through to consumers.

Jeremy Bulow
Director, FTC Bureau of Economics

Jonathan B. Baker
former Director, FTC Bureau of Economics




Robert L. Steiner

In November, 1999, my piece, A Critique of the FTC’s Analysis of the Proposed Tobacco Settlement, appeared on the American Antitrust Institute’s (AAI) web site. Its conclusions were summarized in Bert Foer’s AAI column in the November 21 FTC: WATCH in which Bert provided supporting evidence from his experience as an owner of a chain of jewelry stores. Two FTC Staff Reports had predicted that cigarette wholesalers and retailers would pass through manufacturers’ price increases to consumers without adding a further markup. In November, 1998, cigarette makers announced a large price increase in connection with an expensive settlement with the states. My web site piece presented evidence that the distributive trade had added a substantial $ markup, and it provided a theoretical explanation for this result.

The December 20, 1999 FTC WATCH carried a reply by Jeremy Bulow and Jonathan Baker, the current and former directors of the FTC’s Bureau of Economics. The reply characterized my theoretical framework "as demonstrably incorrect". It did not accept my empirical evidence, yet provided none of its own to support the FTC’s prediction that the pass through to consumers would not exceed 100%. "Without more" Bulow and Baker saw no reason to reconsider the manner in which the FTC analyzes how factory price increases affect consumer prices.

The extent of pass throughs is an important, recurring issue. So I welcome this opportunity to present the "more". It consists of information included in my web site piece that Bulow and Baker have virtually or totally ignored and additional empirical evidence. My approach follows the Marshallian method with its emphasis on the role of induction. Find our what occurred and then provide a coherent explanation. See R. H. Coase, Marshall on Method, Journal of Law & Economics, Vol. XVIII, April 1975.

Accordingly, as my web site piece recounts, I interviewed several wholesalers and several retailers, consulted a veteran executive of a non-lobbying, industry-wide trade association, Darryl Jayson of TMA (Tobacco Merchants Association), and reported the consensus view of tobacco industry analysts of major investment banking firms. All agreed that the distributive trade had in the past, and predicted that it would with the November, 1998, factory price increase, add a substantial further markup. When actual participants and close students of an industry report that something is so, it is highly likely to be so.

To discover whether the trade margin had increased, I used TMA’s careful estimate of manufacturers’ list prices (less only a 2% cash discount) in November, 1998, just before the $.45 list price increase and in August, 1999, just prior to a subsequent price increase. To estimate retail prices I used the Consumer Price Index (CPI) for the same two months. As reported on my web site piece and in the AAI column, I found a significant increase in the $ trade margin (and a slight decrease in the % trade margin). The increased trade margin added $.10 to the retail price per pack of cigarettes before sales tax.

To buffer the consumer shock at the large price increase, many cigarette manufacturers launched an aggressive discounting campaign featuring, according to TMA’s Jayson, "hard coupons". These are physically attached to the goods and the coupon amount is deducted from the price of the cigarettes at checkout. Top officials in the tobacco section of the Consumer Price Index advise that the index reflects any coupons or other discounts offered on or in close proximity to the goods (but does not capture, for instance, newspaper coupons). The retailer is commonly reimbursed for the coupons by the wholesaler who is subsequently reimbursed by the cigarette manufacturer. Since the published manufacturers’ list prices in my calculations did not reflect this reimbursement, the results understate the extent to which the trade marked up the true net manufacturers’ price increases.

Not reported in my web site piece was a second calculation that used the November, 1998, and August, 1999, Producer Price Indexes (PPI) instead of the TMA list prices to represent changes in cigarette manufacturers’ prices. The PPI index picks up all off-invoice promotional discounts given to the original customer. Hence, it rose by less than did the TMA list prices. Using the same pre-price increase trade margin estimated by the FTC, the $ trade margins in the second calculation rose almost $.17 compared to $.10 in the first calculation.

Still, the PPI does not pick up manufacturers’ reimbursements for coupons. Officials at both PPI and CPI concur that the PPI captures far less of the total discounting activity paid for by the cigarette manufacturer than does the CPI! Hence, even my second calculation may understate the extent to which the distributive trade marked up the actual net increase in cigarette manufacturers’ prices.

Thus the quantitative evidence strongly supports the conclusion of participants and of close observers of the cigarette business that the pass through of the factory price increase substantially exceeded 100%. But establishing what occurred is far easier than selecting a theory that explains it. My web site piece identified two forces that would elevate non-invoice marginal costs for tobacco wholesalers and retailers - the sharp rise in shrinkage due to increased theft as the price of cigarettes rise and the increased costs of financing a unit of inventory when its price increases and its rate of turnover declines. I also reported an experienced supermarket executive’s prediction that a major retail price increase would seriously diminish the consumer elasticity of demand for cigarettes. This prediction has apparently come to pass. TMA’s latest estimate shows that in 1999 the elasticity of demand for cigarettes dropped sharply from about (-) 0.4 to about (-) 0.25.

I reasoned that when their non-invoice costs rose and the elasticity of their demand curves fell, wholesalers and retailers would raise their $ margins per unit, although their % margins would slip (Bulow and Baker misquote me here). I suggested that the price increases would be facilitated by "Unfair Sales Laws" in 27 states, since the laws specify minimum allowable margins that increase proportionately with increases in cigarette manufacturers’ prices. While perhaps my approach erred in holding that $ trade margins would rise in the amount specified by the Lerner Index, at least it correctly predicts that margins would rise significantly.

In contrast, Bulow and Baker embrace a model with a monopolistically competitive distribution sector that includes a rather elliptically described "commonly satisfied" condition in which downstream firms pass through less than 100% of a general increase in manufacturers’ prices. Like the proverbial professor who invented a cure for which there was no known disease, this explains a result that did not occur here. Monopolistic competition models contain a zero profit condition that may not be met. Cigarettes show supermarkets an above average return, despite their low % gross margin, due to their rapid turnover and compact size. Michael Lynch, formerly Acting Director of the FTC’s Bureau of Economics, raises a more crucial objection. The less than 100% pass through result rests upon an assumption of diseconomies of scale in distribution. This seems untrue of wholesaling and is clearly untrue of retailing where 100,000 square foot combination discount cum supermarket "superstores", a variety of huge "category killer" retailing formats and massive wholesale clubs have undersold and driven out smaller scale stores.

Yet when returns to scale are assumed, this same Bulow and Baker model predicts a pass through rate greater than 100%, so perhaps it may prove a useful tool after all. Instead of shutting the door on this vital topic, I urge the FTC to adopt the Marshallian method. Accept my empirical findings, or conduct your own investigation and then propose a theory that accounts for what actually occurred.