impact of national generic dairy advertising on dairy markets

Journal of Agricultural and Applied Economics, 29,2(December 1997):303-3130 1997 Southern Agricultural Economics Association

Impact of National Generic DairyAdvertising on Dairy Markets, 1984–95

Harry M. Kaiser

ABSTRACT

The impacts of generic dairy advertising on retail, wholesale, and farm dairy markets areestimated in this study at the national level. The results indicate that generic dairy adver-tising had a major impact on retail, wholesale, and farm markets for the dairy industry.The main conclusion of the study is that farmers are receiving a high return on theirinvestment in generic dairy advertising, i.e., an average rate of return of $3,40 for everydollar invested over the period 1984–95. Moreover, the return on investment in advertisingwas higher in the most recent year, almost double the average for the previous 11 years.

Key Words: dairy, generic advertising, industry econometric model, simulation.

Dairy farmers pay a mandatory assessment of1S@ per hundred pounds of milk marketed inthe continental United States to fund a nationaldemand expansion program. The aims of thisprogram are to increase consumer demand formilk and dairy products, enhance dairy farmrevenue, and reduce the amount of surplusmilk purchased by the government under theDairy Price Support Program. Legislative au-thority for these assessments, which exceed$200 million annually, is contained in theDairy and Tobacco Adjustment Act of 1983.To increase milk and dairy product consump-tion, the National Dairy Promotion and Re-search Board (NDPRB) was established to in-vest in generic dairy advertising andpromotion, nutrition research, education, andnew product development.

The author is an associate professor and director of theCornell Commodity Promotion Research Program inthe Department of Agricultural, Resource, and Mana-gerial Economics at ComelI University.

Funding for this project came from a grant fromthe New York Dairy Promotion Order, and from DairyManagement, Inc. The author thanksJenny Ferrero foreditorial assistance on this manuscript.

The purpose of this study is to estimate theimpacts of the NDPRB generic advertising ef-fort on the U.S. dairy industry. The modelused is based on a dynamic econometric mod-el of the U.S. dairy industry estimated usingquarterly data from 1975 through 1995, and isunique from previous models of the U.S. dairysector in its level of disaggregation. For in-stance, the dairy industry is divided into retail,wholesale, and farm markets, and the retailand wholesale markets separately include fluidmilk, cheese, butter, and frozen products.Econometric results are used to simulate mar-ket conditions with and without the nationalprogram.

The results of this study are important fordairy farmers and policy makers given that thedairy industry has the largest generic promo-tion program of all U.S. agricultural commod-ities. Moreover, since the constitutionality ofsome commodity promotion organizations (in-cluding dairy) is currently being challenged incourt, measurements of the economic impactsof generic advertising are particularly impor-tant at this time. Hence, this study providesinformation that may help the discussion in

304

future legal debatespromotion programs.

Conceptual Model

Journal of Agricultural and Applied Economics, December 1997

concerning commodity

There has been a great deal of research on theimpacts of generic dairy advertising, For ex-ample, in an annotated bibliography of genericcommodity promotion research, Ferrero et al.listed 29 economic studies on dairy over theperiod 1992–96. Some of this research hasbeen at the state level, with New York statebeing studied extensively (e.g., Kinnucan,Chang, and Venkateswaran; Kaiser and Re-berte; Reberte et al.). These studies have usedsingle-equation techniques to estimate demandequations, usually for fluid milk, as functionsof own price, substitute price, income, popu-lation demographics, and advertising. Therehave been several recent national studies con-ducted as well (e.g., Blisard and Blaylock; Liuet al. 1990; Cornick and Cox; Suzuki et al.;Wohlgenant and Clary). Of these analyses, themost disaggregate in terms of markets andproducts is that of Liu et al. (1990), who de-veloped a multiple-market, multiple-productdairy industry model to measure the impactsof fluid milk and manufactured dairy productgeneric advertising.

The econometric model presented here issimilar in structure to the industry model de-veloped by Liu et al. (1990, 1991). FollowingLiu et al. (1990, 1991), the current study em-ploys a partial equilibrium model of the do-mestic dairy sector (with no trade) that dividesthe dairy industry into retail, wholesale, andfarm markets. However, while Liu et al. (1990,199 1) classified all manufactured products intoone category (Class III), the present modeldisaggregates manufactured products intothree classes: frozen products, cheese, and but-ter. This greater degree of product disaggre-gation provides for additional insight into theimpacts of advertising on individual productdemand, e.g., cheese, butter, and frozen prod-uct demand.

In the farm market, Grade A (fluid eligible)milk is produced by farmers and sold towholesalers. The wholesale market is disag-gregate into four submarkets: fluid (bever-

age) milk, frozen products, cheese, and butter.]Wholesalers process the milk into these fourdairy products and sell them to retailers, whothen sell the products to consumers. The mod-el assumes that farmers, wholesalers, and re-tailers behave competitively in the market.This assumption is supported empirically bytwo recent studies. Liu, Sun, and Kaiser esti-mated the market power of fluid milk andmanufacturing milk processors, concludingthat both behaved quite competitively over theperiod 1982–92. Suzuki et al. measured thedegree of market imperfection in the fluidmilk industry and found the degree of imper-fection to be relatively small and decliningover time.

It is assumed that the two major federalprograms that regulate the dairy industry(federal milk marketing orders and the DairyPrice Support Program) are in effect. Since

this is a national model, it is assumed that

there is one federal milk marketing order reg-

ulating all milk marketed in the nation. The

federal milk marketing order program is in-

corporated by restricting the prices wholesal-

ers pay for raw milk to minimum class prices.

For example, fluid milk wholesalers pay the

higher Class I price, while cheese wholesalers

pay the lower Class 111 price.z The Dairy

1All quantities in the model are expressed on amilkfat equivalent (me) basis. Consequently, nonfat,dry milk was not considered in the model.

2Until recently, there was no uniform classificationof raw milk among federal milk marketing orders inthe United States. Some orders utilized a two-classpricing system for raw milk (Class I = beverage milk,Class II = manufactured milk), while others used athree-class pricing system (Class I = beverage milk,Class II = soft manufactured products, and Class III= hard manufacturedproducts). Today, all federal milkmarketing orders utilize four product classes, withClass I being fluid products, Class II being soft dairyproducts, Class III being mostly hard dairy products,and Class IIIa being nonfat, dry milk. The currentmodel assumes a two-class pricing system for rawmilk, which is a valid assumption for the wholesalefluid, cheese, and butter equations in the model. How-ever, the Class H price should be used for the whole-sale frozen product supply equation. Unfortunately, formuch of the 1975–95 period, it is difficult to get anational average Class H price. Moreover, the Class 11price, for most of this period, was usually only mar-ginally higher than the Class III price, ranging from

Kaiser: Generic Dairy Advertising 305

c1Fluid MilkAdvertising

t

RetailFluid

Market

WholesaleFluid

Market

- --

0RetailCheeseMarket 0Retail

ButterMarket

RetailFrozenMarket

1

4

Emis?lmI I I J

A A

TFarm Milk Supply

Figure 1. Conceptual overview of U.S. dairy industry model

Price Support Program is incorporated into

the model by restricting the wholesale cheese

and butter prices to greater than or equal tothe government purchase prices for these

products. With the government offering to

buy unlimited quantities of storable manufac-

tured dairy products at announced purchase

prices, the program indirectly supports the

farm milk price by increasing farm-level milk

demand. A conceptual overview of the model

is presented in figure 1.

Retail markets are defined by sets of supply

and demand functions, in addition to equilib-

rium conditions that require supply and de-

mand to be equal. Since the market is disag-

gregate into fluid milk, frozen products,

cheese, and butter, thereequations, with each setgeneral specification:

(1.1) I?D = f(RPIS”),

(1,2) Rs = j(lw’ls”’),

(1.3) RD = RS = R*,

are four sets of these

having the following

zero to 20c higher than the Class III price. Therefore,this assumption is not much of a departurefrom reality.

where RD and RS are retail demand and sup-

ply, respectively, RP is the retail own price,LVdis a vector of retail demand shifters includ-ing generic advertising, S“ is a vector of retailsupply shifters including the wholesale ownprice, and R* is the equilibrium retail quantity.

The wholesale market is also defined byfour sets of supply and demand functions, andequilibrium conditions. The wholesale fluidmilk and frozen product markets have the fol-lowing general specification:

(2.1) WD = R*,

(2.2) Ws = f(wPls’”’),

(2.3) WS=WD=l@= R*,

where WD and WS are wholesale demand andsupply, respectively, WP is the wholesale ownprice, and S“’ is a vector of wholesale supplyshifters. In the wholesale fluid milk supplyequation, S“’ includes the Class I price, whichis equal to the Class III milk price (i.e., theMinnesota-Wisconsin price) plus a fixed fluidmilk differential. In the frozen products,cheese, and butter wholesale supply functions,SWS includes the Class III price, which is the

306 Journal of Agricultural and Applied Economics, December 1997

most important variable cost to dairy proces-sors. Note that the wholesale level demandfunctions do not have to be estimated since theequilibrium conditions constrain wholesale de-mand to be equal to the equilibrium retailquantity. The assumption that wholesale de-mand equals retail quantity implies a fixed-proportions production technology. Researchby Wohlgenant and Haidacher suggests thismay not be a realistic assumption. Unfortu-nately, the data used as a proxy for nationaldemand are commercial disappearance figureswhich do not distinguish between wholesaleand retail-level demand. Consequently, this as-sumption is necessary.

The direct impacts of the Dairy Price Sup-port Program occur at the wholesale cheeseand butter markets level. It is at this level thatthe Commodity Credit Corporation (CCC)provides an alternative source of demand atannounced purchase prices. In addition, cheeseand butter can be stored as inventories, whichrepresent another source of demand not pres-ent with the other two products. Consequently,the equilibrium conditions for the butter andcheese wholesale markets are different thanthose for the fluid milk and frozen wholesalemarkets. The wholesale cheese and butter mar-kets have the following general specification:

(3.1) WD = R*,

(3.2) Ws = f(wP Is”’),

(3.3) ws = WD + AINV + QSP = QW,

where WD and WS are wholesale demand andsupply, respectively, WP is the wholesale ownprice, SW.’is a vector of wholesale supply shift-ers including the Class III milk price, AZNV ischange in commercial inventories, QSP isquantity of product sold by specialty plants tothe government, and Q“ is the equilibriumwholesale quantity. The variables AZNV and

QSP represent a small proportion of total milkproduction and are assumed to be exogenousin this model.3

~Certain cheese and butter plants sell products tothe government only, regardless of the relationship be-tween the wholesale market price and the purchase

The Dairy Price Support Program is incor-porated into the model by constraining thewholesale cheese and butter prices to be notless than their respective government purchaseprices, i.e.:

(4, 1) WCP z GCP,

(4.2) WBP > GBP,

where WCP and GCP are the wholesalecheese price and government purchase pricefor cheese, respectively, and WBP and GBP

are the wholesale butter price and government

purchase price for butter, respectively.

Because of the Dairy Price Support Pro-

gram, four regimes are possible: (a) WCP >

GCP and WBP > GBP, (b) WCP > GCP andWBP = GBP, (c) WCP = GCP and WBP >

GBP, or (d) WCP = GCP and WBP = GBP.In the cheese and butter markets, specific ver-sions of equilibrium condition (3.3) are appli-cable to the first regime, which is the com-petitive case. In the second case, where thecheese market is competitive but the buttermarket is not, the wholesale butter price is setequal to the government purchase price forbutter and the equilibrium condition ischanged to

(3.3b) WBS = WBD + AINv, + QSP,

+ GB = WB,

where GB is government purchases of butter,

which becomes the new endogenous variable

replacing the wholesale butter price. For the

third case, where the butter market is compet-

itive but the cheese market is not, the whole-

price. These are general balancing plants that removeexcess milk from the market when supply is greaterthan demand, and process the milk into cheese andbutter which is then sold to the government. Becauseof this, the quantity of milk purchased by the gover-nmentwas disaggregated into purchases from these spe-cialized plants and other purchases. In a competitiveregime, the “other purchases” are expected to be zero,while the purchases from specialty plants may be pos-itive. The QSPCand QSPb variables were determinedby computing the average amount of government pur-chases of cheese and butterduring competitive periods,i.e., when the wholesale price was greater than the pur-chase price for these two products.


sale cheese price is set equal to the govern-ment purchase price for cheese and theequilibrium condition is changed to

(3.3c) WCS = WCD + AINVC +- QSPC

+ GC = WC,

where GC is government purchases of cheese,which becomes the new endogenous variablereplacing the wholesale cheese price. Finally,for the last case where both the cheese and thebutter markets are not competitive, the whole-sale cheese and butter prices are set equal totheir respective government purchase pricesand the equilibrium conditions are changed to(3.3b) and (3.3c).4

The farm raw milk market is representedby the following milk supply equation:

I fixed fluid milk differential (therefore theClass I price is equal to F’111 + d), WFS iswholesale fluid milk supply, WFZS is whole-sale frozen product supply, WCS is wholesalecheese supply, and WBS is wholesale buttersupply.

Finally, the model is closed by the follow-ing equilibrium condition:

(5.3) FMS = WFS + WFZS + WCS + WBS

i- FUSE + OTHER,

where FUSE is on-farm use of milk and OTH-

ER is milk used in dairy products other thanfluid milk, frozen products, butter, and cheese.Both of these variables represented a smallshare of total milk production and were treatedas exogenous.

Econometric Estimation(5. 1) FMS = f(E[AMP] I S~m),

where Fk2S is commercial milk marketing inthe United States, E[AMP] is the expected allmilk price, and S@ is a vector of milk supplyshifters. As in the model developed by La-France and de Gorter, and by Kaiser, a perfectforesight specification is used for the expectedfarm milk price.

The farm milk price is a weighted average7of the class prices for milk, with the weightsequal to the utilization of milk among prod-ucts :

(5.2) AMP = [(PIII + d) X WFS + PIII

x (WFZS + WCS + WBS)]

+ [WFS + WFZS + WCS + WBS],

where PIII is the Class 111price, d is the Class

4Because the marketstructureis different undereach of these four regimes, using conventional two-stage least squares to estimate equations (1.1) through(4.2) may result in selectivity bias. Theoretically, aswitching simultaneous system regression procedureshould be applied, which is described in Liu et al.(1990, 1991). However, this procedure was not usedhere because it was beyond the scope of this project.Applying this procedure to the level of disaggregationof this model’s manufactured product market wouldhave been extremely cumbersome, and the costs of do-ing so were judged to be greater than the potentialbenefits,

The equations were estimated simultaneouslyusing two-stage least squares and quarterlydata from 1975 through the third quarter of1995. The econometric package used wasEViews (Hall, Lilien, and Johnston). Allequations in the model were specified in dou-ble-logarithm functional form. In terms ofstatistical fit, most of the estimated equationswere found to be reasonable with respect toR2. In all but two equations, the adjusted co-efficient of determination was above 0.88.The two equations that were the most difficultto estimate were the retail butter demand andsupply equations, which had the lowest R*values (O.47 and 0.55, respectively). Estima-tion results are available from the authorupon request.

The retail market demand functions wereestimated on a per capita basis. Retail demandfor each product was specified to be a functionof the following variables: (a) retail productprice, (b) price of substitutes, (c) per capitadisposable income, (d) quarterly dummy vari-ables to account for seasonal demand, (e) atime trend variable to capture changes in con-sumer tastes and preferences over time,5 and

s Several functional forms were specified for thetime trend, including linear, log-linear, and exponentialforms, The form yielding the best statisticalresultswaschosen for each equation.


(~) genetic advertising expenditures to mea-sure the impact of advertising on retail de-mand. In all demand functions, own prices andincome were deflated by a substitute productprice index. This specification was followedbecause there was strong correlation betweenthe substitute price and own price for eachdairy product. The consumer price index fornonalcoholic beverages was used as the sub-stitute price in the fluid milk demand equation,the consumer price index for meat was usedas the substitute price in the cheese demandequation, the consumer price index for fat wasused as the substitute price in the butter de-mand equation, and the consumer price indexfor food was used as the substitute price in thefrozen product demand equations. To measurethe advertising effort of the NDPRB, genericadvertising expenditures for fluid milk andcheese were included as explanatory variablesin the two respective demand equations. GGe-neric advertising expenditures for butter andfrozen products were not included for two rea-sons. First, the NDPRB has not invested muchmoney into advertising these two products,Second, including generic butter and frozenproduct advertising expenditures in an earlierversion of the model resulted in highly statis-tically insignificant estimated coefficients.Branded advertising expenditures were initial-ly included in the estimation, but were omittedfrom the final regression because the coeffi-cients were highly statistically insignificant.

To capture the dynamics of advertising, ge-neric advertising expenditures were specifiedas a second-order polynomial distributed lagwith both endpoint restrictions imposed.Based on previous research (e.g., Liu et al.199 1; Kaiser), a lag length of four quarterswas chosen. Finally, a first-order moving av-erage error structure was imposed on the retailfluid milk demand equation and a first-orderautoregressive error structure was imposed onthe retail cheese demand equation to correctfor autocorrelation.

Based on the econometric estimation, ge-neric fluid milk advertising had the largest

6All generic advertising expenditures came fromvarious issues of Leading National Advertisers.

long-run advertising elasticity of 0.021, andwas statistically different from zero at the 190significance level.’ This means a 1?ZO increasein generic fluid advertising expenditures re-

sulted in a 0.021 !ZOincrease in fluid demandon average over this period, which is compa-rable to the results of previous studies. Forexample, Liu et al. (1990) estimated a long-run fluid milk advertising elasticity of 0.0139using national data; Kinnucan (1986) estimat-ed a long-run fluid milk advertising elasticityof 0.051 for New York City; and Kinnucan,Chang, and Venkateswaran estimated a long-run fluid milk advertising elasticity of 0.016for New York City. Generic cheese advertisingwas also positive and statistically significantfrom zero at the 170 significance level and hada long-run advertising elasticity of 0.016.

The retail supply for each product was es-timated as a function of the following vari-ables: (a) retail price; (b) wholesale price,which represents the major variable cost to re-tailers; (c) producer price index for fuel andenergy; (d) average hourly wage in the foodmanufacturing sector; (e) time trend variable;

(f) qutierly dummy variables; and (g) laggedretail supply. The producer price index for fueland energy was used as a proxy for variableenergy costs, while the average hourly wagewas used to capture labor costs in the retailsupply functions. All prices and costs were de-flated by the wholesale product price associ-ated with each equation. The quarterly dummyvariables were included to capture seasonalityin retail supply, while the lagged supply vari-ables were incorporated to represent capacityconstraints. The time trend variable was in-cluded as a proxy for technological change inretailing. Not all of these variables remainedin each of the final estimated retail supplyequations due to statistical significance artd/orwrong sign on ‘the coefficient. Finally, a first-order moving average error structure was im-

7The long-run advertising elasticity was computedby simulating a 197.permanent increase in advertisingexpenditures for the period 1984–95, which was usedwith the baseline simulation results for actual historicaladvertising expenditures to determine the elasticity.


posed on the retail cheese and frozen productsupply equations.

The wholesale supply for each product wasestimated as a function of the following vari-ables: (a) wholesale price; (b) the appropriateclass price for milk, which represents the mainvariable cost to wholesalers; (c) producer priceindex for fuel and energy; (d) average hourlywage in the food manufacturing sector; (e)time trend variable; (~) quarterly dummy vari-ables; (g) lagged wholesale supply; and (h)two dummy variables for the cheese and butterdemand functions corresponding to the MilkDiversion Program and the Dairy TerminationProgram, which were two supply control pro-grams implemented over some of this period.The producer price index for fuel and energywas included because energy costs are impor-tant variable costs to wholesalers, while theaverage hourly wage was used to capture laborcosts in the wholesale supply functions. Allprices and costs were deflated by the price offarm milk, i.e., class price. The quarterly dum-my variables were used to capture seasonalityin wholesale supply, lagged wholesale supplywas included to reflect capacity constraints,and the trend variable was incorporated as ameasure of technological change in dairyproduct processing. Not all of these variablesremained in each of the final estimated whole-sale supply equations due to statistical signif-icance and/or wrong sign on the coefficient.Finally, a first-order moving average errorstructure was imposed on the wholesale fluidmilk and frozen product supply equations.

For the farm milk market, the farm milksupply was estimated as a function of the fol-lowing variables: (a) ratio of the farm milkprice to feed price (16% protein content), (b)ratio of the price of slaughter cows to the feedprice, (c) lagged milk supply, (d) interceptdummy variables to account for the quartersthat the Milk Diversion and Dairy TerminationPrograms were in effect, (e) a dummy variablefor the second quarter, and (~) time trend vari-able. The 16% protein feed price representsthe most important variable costs in milk pro-duction, while the price of slaughtered cowsrepresents an important opportunity cost todairy farmers. Lagged milk supply was in-

cluded as biological capacity constraints tocurrent milk supply.

Market Impacts of the NDPRB

To examine the impacts that the NDPRB hadon the market over the period 1984.3–1 995.3,the model was simulated under two scenariosbased on generic advertising expenditures: (a)

a historic scenario, where advertising levelswere equal to actual generic advertising ex-penditures, and (b) a no-NDPRB scenario,where quarterly values of generic advertisingexpenditures were equal to quarterly levels forthe year prior to the adoption of the NDPRB,i.e., 1983.3–1984.2. A comparison of thesetwo scenarios provides a measure of the im-pacts of the NDPRB on the dairy markets. Ta-ble 1 presents the quarterly averages of priceand quantity variables for two time periods:1984.3–1995.3, and 1994.3–1995.3. The lasttwo columns in the table give the percentagechange in each variable due to the NDPRB forthe life of the program and the most recentyear, respectively. The results for the longertime period are discussed first.

It is clear from these results that theNDPRB had an impact on the dairy marketfor the period 1984.3–1995.3. For example,the generic advertising effort of the NDPRBresulted in a 0.91% increase in fluid sales anda 5.62% increase in retail fluid price comparedto what would have occurred in the absenceof this national program, Note that since theown price elasticity of fluid milk demand wasestimated to be quite inelastic (– O.12), themodest increase in fluid sales due to advertis-ing caused a sizable increase in price. The in-crease in fluid sales also caused the wholesalefluid price to increase by 4. 10?ZO.The increasein advertising expenditures due to the NDPRBalso had positive impacts on the retail cheesemarket. Retail cheese quantity and price were0.48910and 0.8 1% higher, respectively. The in-crease in cheese sales caused the wholesalecheese price to rise by 2.75910.

Although generic butter and frozen productadvertising were not included in the retail de-mand equations, generic fluid milk and cheeseadvertising by the NDPRB had some indirect,


Table 1. Simulated Quarterly Values for Market Variables With and Without NDPRB, Aver-aged over 1984.3–1995.3, and over 1994.3 –1995.3

1994.3-1984,3 –1995.3 Average 1995.3

With Without Avg. ~0 Avg. VoMarket Variables Units NDPRB NDPRB Change Change

Fluid demandlsupplyFrozen demandlsupplyCheese demandCheese supplyButter demandButter supplyTotal demandRetail fluid priceRetail frozen priceRetail cheese priceRetail butter priceWholesale fluid priceWholesale frozen priceWholesale cheese priceWholesale butter priceClass III priceAll milk priceCCC cheese purchasesCCC butter purchasesCCC purchasesMilk supplyProducer surplus

bil. lbs. me’bil. lbs. mebil. lbs. mebil. lbs. mebil. lbs. mebil. Ibs. mebil. lbs. me1982–84 = 1001982–84 = 1001982–84 = 1001982–84 = 1001982 = 1001982 = 100$ilb.$/lb.$Icwt$Ycwtbil. lbs. mebil. lbs. mebil. lbs. mebil, lbs.bil. $

13.593.16

12.2612.315.246.65

34.25114.94130.11123.3895.41

124.94126.59

2.461.10

12.6513.600.061.421,48

36.524.63

13.473.16

12,2012.265.246,67

34.07108.47129,16122.3895.31

119.82125.23

2.401.09

12.3913.340.071.441.51

36.384.52

0.91–0,16

0.480.42

–0.06–0.39

0,515.620.730,810.104,101.072.750.272.031.93

–12.00–1,59–2.01

0.402,34

1.72–0.24

0.270.27

–0.21–0.47

0,7010.11

1.100.490.396.741.621.781,123.042.930.00

–2.25–2.25

0.623.52

“ The notation “me” standsfor milk equivalent.

but minor impacts on butter and frozen prod-uct markets. For example, the retail andwholesale frozen product price increased, onaverage, by 0.73% and 1.07Y0, respectively,due to the NDPRB advertising effort. The in-creases in frozen product prices were primarilydue to the higher Class III milk price that man-ufacturers had to pay under the NDPRB ad-vertising scenario. Advertising by the NDPRBhad little impact on retail and wholesale butterprices, but butter supply declined by 0.399i0under NDPRB advertising. The decline in but-ter supply was due to a higher average ClassIII price.

The NDPRB also had an impact on pur-chases of cheese and butter by the govern-ment. The modest increase in cheese demandrelative to the increase in wholesale supplydue to NDPRB advertising caused cheese pur-chases by the government to fall by 12Y0, on

average, over this period. Likewise, while but-ter demand did not change, the 0.39% de-crease in butter supply due to generic adver-tising by the NDPRB caused butter purchasesby the government to decrease by 1.59% overthe period. Total dairy product purchases bythe government were 2.0 1% lower in theNDPRB scenario.

The introduction of the NDPRB also hadan impact on the farm market over the previ-ous 11 years. The Class III and farm milk pric-es increased by 2.039i0 and 1.9370 under thenational program due to an increase of 0.5170in total milk demand. Farm supply, in turn,increased by 0.40Y0. Farmers were better offunder the NDPRB since producer surplus av-eraged 2.34% higher with the program. Onebottom-line measure of the net benefits of theNDPRB to farmers is the rate of return, whichgives the ratio of benefits to costs of the na-


tional program. Specifically, this rate-of-returnmeasure was calculated as the change in pro-ducer surplus, due to the NDPRB, divided bythe costs of funding this program. The cost ofthe program was measured as the 15@ per hun-dredweight assessment times total milk mar-keting. In the year prior to the program, farm-ers voluntarily contributed 6.3@ per hundredweight. Therefore, the difference in cost dueto the national checkoff was assumed to be thedifference between 0.0015 times milk market-ing (in billion pounds) under the NDPRBscenario minus 0.00063 times milk marketingin the no-NDPRB scenario. The resultsshowed that the rate of return from theNDPRB was 3.40 over the 11-year period.This means that an additional dollar investedin generic advertising would return $3.40 inproducer surplus to farmers. The farm-levelrate of return was lower than estimates of 4.77by Liu et al. (1990) for the period 1975. l–1987.4, 4.60 by Kaiser and Forker for the pe-riod 1975.1-1990.4, and 5.40 for the period1975. 1–1993.4 by Kaiser. A 95% confidenceinterval was calculated for the rate-of-returnestimate by simulating the two scenarios sep-arately with the lower and upper limits of the95% confidence interval for the fluid milk andcheese advertising coefficients in the retail de-mand equations. The lower and upper limitsof the 95% confidence interval for the rate ofreturn were 0.60 and 7.93, respectively.

The last column in table 1 gives the im-pacts of the NDPRB for the most recent year.In general, the most recent year’s results dem-onstrate larger market impacts of the NDPRBadvertising effort than the 11-year average.For example, generic advertising by theNDPRB in the last four quarters resulted in a1.72% increase in fluid sales and a 10.11 Yoincrease in retail fluid price relative to whatwould have occurred without NDPRB adver-tising. The increase in fluid sales caused thewholesale fluid price to increase by 6.74?Z0,onaverage, over the last year in the simulation.One reason for larger NDPRB advertising im-pacts on the fluid market in the recent periodis due to greater emphasis on fluid advertisingin recent years.

While the advertising effort of the NDPRB

in the last year continued to have a positiveimpact on the cheese market, the magnitude ofimpacts was smaller than the overall averagefor the 1l-year period. Again, this was due tothe increase in fluid milk advertising and de-crease in cheese advertising in recent years.Retail cheese sales and prices were 0,27% and0.49% higher, respectively, due to NDPRB ad-vertising. The modest increase in cheese salesresulted in an average increase of 1.78% in thewholesale cheese price as well.

The most recent year’s advertising impactson government purchases were smaller thanthe 11-year average because purchases were atmuch smaller levels to begin with. For ex-ample, purchases of cheese by the governmentwere predicted to be zero both with and with-out NDPRB advertising. Butter purchases,however, were 2.2590 lower with advertising.

The farm market impacts due to theNDPRB were larger in the most recent yearthan the 11-year period. The Class 111 andfarm milk price increased by 3.04% and2.93%, respectively, on average for the pastyear due to the national program. This was dueto an increase in total milk demand of 0.70%because of generic advertising. Farm supply,in turn, was 0.62?Z0higher in the NDPRB sce-nario. Farmers were better off under theNDPRB since producer surplus was 3.52%higher, and the rate of return was 6.43. Thelower and upper limits of the 95’%0confidenceinterval for the rate of return were 1.44 and13.74, respectively. Note that the rate of returnwas almost twice as high for the most recentyear than it was, on average, for the past 11years. Therefore, the results suggest that thenet benefits of the NDPRB to farmers havebecome larger in recent years.

Conclusion

The purpose of this study was to analyze theimpacts of generic dairy advertising by theNational Dairy Promotion and Research Boardon retail, wholesale, and farm dairy markets.The results indicated that the NDPRB had amajor impact on retail, wholesale, and farmmarkets for the dairy industry. The main con-clusion of the study is that farmers are receiv-


ing a high return on their investment in ge-

neric dairy advertising. Moreover, the return

on investment in advertising has been higher

in the most recent year than the average for

the previous 11 years. The higher profitability

of generic advertising in recent years may be

attributed to increases in efficiency by pro-

gram managers due to experience in running

the program.

Given the current legal debate over man-

datory commodity checkoff programs, the ev-

idence from this study can be used to dem-

onstrate that generic advertising does have a

significant impact on the market. The impacts

of advertising tend to be more profound in in-

creasing price than quantity, which is due tothe inelastic nature of demand for milk andcheese, These estimated impacts need to becompared with other options producers havefor marketing their product (e.g., nonadvertis-ing promotion, research, new product devel-opment, etc.) in order to determine the opti-mality of the current investment ofadvertising. Consequently, these results shouldbe viewed as a first step in the evaluation pro-cess,

While there are advantages to the industrymodel used in this study, there are also severalshortcomings that need to be pointed out.First, the advertising impacts may be over-stated due to the assumption of fixed propor-tions. As Kinnucan (1997) points out, thefixed proportions assumption does not allowfor input substitution, which may cause de-rived-demand elasticities for farm output to beunderstated and profits from advertising to beoverstated. Second, the interaction of relatedcommodities in the retail demand equations issomewhat limited, especially compared withother econometric techniques, particularly de-mand system models. Hence, cross-commod-ity effects are not measured. Finally, the mod-el did not include several other activities ofthe NDPRB such as nonadvertising promotionand research. While advertising is by far thelargest investment by the NDPRB, these otheractivities may also have an impact on demandfor milk and dairy products. Unfortunately,these data could not be obtained for this study.

There are two directions that could be use-

ful for future research. Obviously, inclusion ofother marketing activities by the NDPRBwould be useful because then the model couldbe used to determine the optimal allocation ofdairy farmer checkoff funds across marketingactivities. In addition, spatial disaggregation ofthe model into several regions of the UnitedStates, particularly for fluid milk, would bevaluable. Although manufactured dairy prod-ucts are well represented as a national market,fluid milk markets tend to be regional inscope, and fluid milk marketing orders causedifferent price surfaces for fluid milk. Region-al disaggregation of fluid milk markets wouldalso make the model a valuable tool in ex-amining dairy policy questions on such issuesas federal milk marketing order consolidation.

References

Blisard, N., and J.R. Blaylock. “A Double-HurdleApproach to Advertising: The Case of Cheese. ”Agribus.: An Internat. J. 8(1992): 109–20.

Cornick, J., and T.L. Cox. “Endogenous SwitchingSystems: Issues, Options, and Application tothe U.S. Dairy Sector. ” J. Agr. Econ. Res.

44(1994):28-39.

Ferrero, J., L. Boon, H.M. Kaiser, and O.D. Forker.“Annotated Bibliography of Generic Commod-ity Promotion Research. ” Revised. NICPRERes. Bull. No. 96-3, National Institute for Com-modity Promotion Research and Evaluation,Dept. of Agr., Resour., and Managerial Econ.,Cornell University, 1996.

Hall, R. E., D.M. Lilien, and J. Johnston. EViews

Users Guide, Version 1.0. Irvine CA: Quanti-tative Micro Software, 1994.

Kaiser, H.M. “An Analysis of Generic Dairy Pro-motion in the United States.” Pub. No. NICPRE95-01, Dept. of Agr., Resour., and ManagerialEcon., Cornell University, 1995.

Kaiser, H.M., and O.D. Forker, “Analysis of Ge-neric Dairy Advertising Scenarios on Retail,Wholesale, and Farm Milk Markets. ” A.E. Res.Pub. No. 93-3, Dept. of Agr., Resour., and Man-agerial Econ., Cornell University, 1993.

Kaiser, H.M., and J.C. Reberte. “Impact of GenericFluid Milk Advertising on Whole, Lowfat, andSkim Milk Demand.” J. Dairy Sci. 79(1996):2284–91.

Kinnucan, H.W. “Demographic Versus Media Ad-vertising Effects on Milk Demand: The Case of


the New York City Market.” Northeast. J. Agr.

.&on. 15(1986):66–76.

—. “Middlemen Behavior and Generic Ad-vertising Rents in Competitive, Interrelated In-dustries. ” Austral. J. Agr. and Resour. Econ.41(1997, forthcoming).

Kinnucan, H,W., H.S. Chang, and M. Venkates-waran. “Generic Advertising Wearout.” Rev.Mktg. and Agr. Econ. 61( 1993):401–16.

LaFrance, J.T., and H. de Gorter. “Regulation in aDynamic Market: The U.S. Dairy Industry. ”Amer. J. Agr. Econ. 67(1985):821–32.

Leading National Advertisers, Inc. Leading Nation-

al Advertisers and Class/Brand QTR $. New

York, various issues, 1975–95.Liu, D. J., H.M. Kaiser, O.D. Forker, and T.D.

Mount. “An Economic Analysis of the U.S. Ge-neric Dairy Advertising Program Using an In-dustry Model. ” Northeast. J. Agr. and Resour.

Econ. 19(1990):37–48.

Liu, D.J., H.M. Kaiser, T.D. Mount, and O.D. Fork-er. “Modeling the U.S. Dairy Sector with Gov-ernment Intervention. ” West. J. Agr. Econ.16(1991):360–73.

Liu, D. J., C.-H. Sun, and H.M. Kaiser. “MarketConduct Under Government Price Interventionin the U.S. Dairy Industry.” J. Agr. and Resour.Econ. 20(1995):301–15.

Reberte, J.C., H.M. Kaiser, J.E. Lenz, and O.D.Forker. “Generic Advertising Wearout: TheCase of the New York City Fluid Milk Cam-paign.” J. Agr. and Resour. Econ. 21(1996):199–209.

Suzuki, N., H.M, Kaiser, J.E. Lenz, and O.D. Fork-er. “Evaluating U.S. Generic Milk AdvertisingEffectiveness Using an Imperfect CompetitionModel. ” A.E. Res. Pub. No. 93-8, Dept. of Agr.Econ., Cornell University, 1993.

Wohlgenant, M. K., and C.R. Clary. “Developmentand Measurement of Farm-to-Retail Price Link-age for Evaluating Dairy Advertising Effective-ness. ” J. Agr. Econ. Res. 44,4(1994): 18–28.

Wohlgenant, M. K., and R.C. Haidacher. “Retail-to-Farm Linkage for a Complete Demand Systemof Food Commodities. ” Pub. No. TB- 1775,USDA/Economic Research Service, Washing-ton DC, 1989.

impact of national generic dairy advertising on dairy markets

Documents