8/13/2019 Proiect Masini Articol

1/5

The Journal of Socio-Economics 41 (2012) 2630

Contents lists available at SciVerse ScienceDirect

The Journal of Socio-Economics

j ou rna l ho me pag e : www.e l sev i e r. com/ loca t e / soceco

What affects customer success when bargaining for a new car? Some empiricalevidence

Ofer H. Azar

Department of Business Administration, Guilford Glazer Faculty of Business and Management, Ben-Gurion University of the Negev, P.O.B.653, Beer-Sheva 84105,Israel

a

r

t

i

c

l

e

i

n

f

o

Article history:Received 12 August 2010Received in revised form 20 August 2011Accepted 18 October 2011

JEL classication:L81C78D40L92M31

Keywords:Car dealersBargainingNegotiationsAutomobile industryConsumer behavior

a

b

s

t

r

a

c

t

Data about 233 new car models were collected, and a measure of customer success in bargaining for anewcar (alpha) was created by computing the ratio between the discount received on the manufacturerssuggested retail price (MSRP) and the negotiable range (MSRP dealers car cost). One hypothesis wasthat customers who purchase more expensive cars succeed less in bargaining because of their higher timevalue. A second hypothesis was that a positive correlation between the negotiable range and alpha shouldexist, because of either customer incentives to bargain or dealers bargaining strategy. Both hypotheseswere supported by the data.

2011 Elsevier Inc. All rights reserved.

1. Introduction

In many situations involving two or more players, outcomes aredetermined by negotiations and bargaining among the involvedparties. This happens not just in economic and business situations,but also in social, political, and diplomatic situations. The impor-tance of negotiations and bargaining in these various situationshas led to a large literature in various elds, including economics,psychology, and conict management. Much of this literature usestheoretical models (e.g., Rubinstein, 1982 ), lab experiments (e.g.,Moran and Ritov, 2002; Gneezy et al., 2003 ), or eld experiments(e.g., List, 2004 ).

One market in which most transactions are conducted after abargaining process is the US market for new cars. This is an inter-esting market to explore in order to obtain empirical evidenceabout bargaining outcomes for several reasons: rst, the amountsof money involved are large, so customers and dealers have large

I thank http://www.Edmunds.com for the permission to use the data down-loaded from their website. Financial support from thePhillipe Monaster Center forEconomic Research at Ben-Gurion University of the Negev is gratefully acknowl-edged.

Tel.: +972 8 6472675; fax: +972 8 6477691.E-mail address: azar@som.bgu.ac.il

incentives to bargain and negotiate optimally; the effect of doingso can be savings (or extra prots) of hundreds andeven thousandsof dollars. Consequently, people indeed take bargaining for a newcar seriously, and various websites and even books give advice onhow to negotiate effectively for a new car (e.g., Lyle, 1999; Bragg,2004; Ford, 2005; Reed, 2005 ).

Second, bargaining for a new car is a process that can take manyhours and even days, 1 so also on the effort part choosing how andhow much to negotiate has signicant implications. In addition,many cars are sold each year, various data about the cars and thetransactions are available, the same good is sold to many differentpeople (as opposed to some other markets involving large transac-

tions such as the real estate market), and the goods value is notvery subjective (as in the market for art, for example). Becauseof these attractive characteristics of the car market, various stud-ies have used this market to explore different issues. Ayres andSiegelman (1995) and Goldberg (1996) , for example, examine theissue of discrimination in the car market. Goff et al. (1997) studyhow salesperson behavior affects customer satisfaction with the

1 Someof theeffort associated withcar purchasebargainingis to elicitoffersfromcompeting dealerships, and eliciting several offers when each might require a fewhours to getto thedealership and negotiate caneasily take several days.

1053-5357/$ see front matter 2011 Elsevier Inc. All rights reserved.

doi: 10.1016/j.socec.2011.10.007

http://localhost/var/www/apps/conversion/tmp/scratch_2/dx.doi.org/10.1016/j.socec.2011.10.007http://localhost/var/www/apps/conversion/tmp/scratch_2/dx.doi.org/10.1016/j.socec.2011.10.007http://www.sciencedirect.com/science/journal/10535357http://www.elsevier.com/locate/socecohttp://www.edmunds.com/mailto:azar@som.bgu.ac.ilhttp://localhost/var/www/apps/conversion/tmp/scratch_2/dx.doi.org/10.1016/j.socec.2011.10.007http://localhost/var/www/apps/conversion/tmp/scratch_2/dx.doi.org/10.1016/j.socec.2011.10.007mailto:azar@som.bgu.ac.ilhttp://www.edmunds.com/http://www.elsevier.com/locate/socecohttp://www.sciencedirect.com/science/journal/10535357http://localhost/var/www/apps/conversion/tmp/scratch_2/dx.doi.org/10.1016/j.socec.2011.10.007

8/13/2019 Proiect Masini Articol

2/5

O.H. Azar / The Journal of Socio-Economics 41 (2012) 2630 27

salesperson, dealer, product and manufacturer in a sample of newcar buyers. Busse et al. (2006) examine to what extent promotionsoffered by auto-manufacturers reduce the net price paid by cus-tomers. Verhoef et al. (2007) study brand and dealer retention inthe new car market.

This article examines the relationship between the outcomeof the bargaining (between the customer and the dealer) and thecharacteristics of the car purchased, such as the cars MSRP (man-ufacturers suggested retail price) and the dealers cost of the car.Of particular interest is the question what affects the success of the customer in the bargaining process. The analysis supports thehypothesis that higher time value of the customer leads to less suc-cessin bargaining.The analysisalsoshows thatwhen thenegotiableprice range is higher, the customer secures a larger relative portionof it, suggesting that customers make more bargaining effort whenthey have more to gain from bargaining. However, another poten-tialexplanation for the positive correlation between the negotiableprice range and the customers success is also suggested. The restof the article is organized as follows: Section 2 discusses the datacollected, Section 3 presents two hypotheses to be tested using thedata, Section 4 presents the results of the data analysis, andthe lastsection concludes.

2. The data

The data were collected in 2003 by recording from the web-site http://www.Edmunds.com various characteristics about newcars. For each car that appeared in their database, only one trimlevel was chosen (since various trim levels of the same car maybe very similar to each other) , and various data about the carwere recorded. The data collected included the cars MSRP, theinvoice price (the price that the dealer pays the manufacturer forthe car), the dealer holdback (a sum of money that the manufac-turer pays thedealer back on thecar), andthe National True MarketValue (the average price that consumers payfor this car, computedby http://www.Edmunds.com based on actual transactions data,henceforth NTMV). In some cases the available data about a carwere insufcient, for example because http://www.Edmunds.comdid not have information about the invoice price. In total, useabledata were obtained about 233 different models.

The invoice price minus the dealer holdback is the net cost of the car for the dealer, and is therefore dened below as the dealerscost. Since the dealer cannot make a prot by selling cars belowtheir cost, the dealers cost represents a lower bound on the nego-tiated price of the car. The MSRP is the price that the manufacturersuggests for a car, and it is also known as the sticker price becausethis is the price that is written on the car. However, customers gen-erally know that this price is only the starting point, and that it ispossible to negotiate with the dealer about the price. Therefore theMSRP can be interpreted as the upper bound to how much a cus-

tomer might pay for the car. Consequently, the range between thedealers cost and the MSRP is the range in which bargaining takesplace. 2 Henceforth, letus use Costfor thedealerscost, anddenethe term negotiable range by:

Negotiable range = MSRP Cost.

2 The dealer obviously knows both his cost and the MSRP. The customer knowsthe MSRP, and can easily obtain the information about the dealers cost, since theinformation about the invoice priceand the dealer holdback is publiclyavailable forfree in many websites, such as Edmunds.com. Even if the customer does not knowthe dealers cost, the dealers bargaining behavior mightprovide signals about it. Inany case, none of the analysis below depends on whether the customer knows the

dealers cost or not.

We cannow use theratiobetween thediscountthat theaveragecustomer obtained on the MSRP (which is equal to MSRP NTMV)and the total negotiable range (MSRP Cost) as a measure of thecustomers success in thebargaining process. Letus dene this ratioas :

=MSRP NTMVMSRP Cost

Notice that is a number between 0 and 1, where a higher rep-resents more customers success in the bargaining process.

Table 1 presents summary statistics of the data. We can see thatforthe average carin thesample,the negotiable range,whichis alsothe hypothetical markup in dollars (i.e., how much above cost thedealer would sell the car ifhe could sellit for the price suggested bythe manufacturer), is $3741, or 12.9% of the cars cost to the dealer.The actual markup (according to average real transaction data, andreecting the outcome of the bargaining between the dealer andthe customer), which is given by NTMV Cost, is on average $1851(or 5.7% of the cars cost); the median markup, however, is muchsmaller (since the average is affected by some high markups onexpensive cars) only $1172, or 4.5% of the cars cost. On average,the customer receives after the bargaining process a discount of $1890 on the MSRP, or 54.5% of the negotiable range.

3. Hypotheses

The most interesting aspect that the data allow to examine isthe relationship between the customers success in the bargaining( ), the cars price, and the negotiable range. Two main hypothe-ses seem interesting to explore. One hypothesis is that customerswith lower value of time will have more success in the bargainingprocess. The reason is that thebargaining process requires time, forexample, staying longer in the dealership in order to keep negoti-ating about thecars price,spending more time eliciting offers fromother dealers in order to be able to lower the negotiated price, etc.Assuming that there are diminishing marginal returns to effort inbargaining (i.e., that the rst hour devoted to bargaining reducesthe price by more than the fth hour), the optimal strategy for thecustomer is to bargain as long as the marginal returns of additionalbargaining areabove thevalueof thecustomers time, andstopbar-gaining once this condition no longer holds. The lower the value of thecustomers time, themore bargaining is optimal, andthe higher should be.

The value of the customers time is closely related to the cus-tomers income, for two reasons. First, many people can earn moreby working additional hours. By spending time doing things otherthan working (such as bargaining on thecar andeliciting additionalprice quotes from other dealers), they forgo this additional earningopportunity. Consequently, the higher is the hourly income of theperson, the more he forgoes by dedicating an additional hour tobargaining. Second, people withhigher income are generally richer,

and richer people have on average a higher willingness to pay formost things, including leisure time. Spending less time on bargain-ing increases the leisure time the person has for more enjoyableactivities. For example, in a situation where additional two hoursof bargaining effort are likely to result in saving an additional $100,it is more likely that someone with a mortgage and a credit carddebt will continue bargaining than that someone with two mil-lion dollars in the bank will do the same. These two considerationsimply that a higher income leads on average to a higher value of time.

In addition, a higher income also leads on average to buy moreexpensive cars, because the person has larger nancial resourcesand thus can afford to spend more money on the various goodsand services he consumes, including cars. Because a higher income

leads on average to both a higher value of time and to buying more

http://www.edmunds.com/http://www.edmunds.com/http://www.edmunds.com/http://www.edmunds.com/http://www.edmunds.com/http://www.edmunds.com/

8/13/2019 Proiect Masini Articol

3/5

8/13/2019 Proiect Masini Articol

4/5

O.H. Azar / The Journal of Socio-Economics 41 (2012) 2630 29

Table 2Means of various variables by MSRP group.

MSRP rangeVariable

Below $20,000 $20,000$25,000 $25,000$30,000 $30,000$35,000 $35,000$45,000 Above $45,000

MSRP 17,320 22,512 27,535 32,732 39,070 65,653Invoice 16,094 20,570 25,061 29,449 35,292 59,277NTMV 16,408 20,921 25,599 30,196 36,717 63,446Dealer cost 15,671 19,959 24,424 28,595 34,406 58,195MSRP Cost ($) 1649 2552 3111 4137 4664 7459

(MSRP Cost)/Cost 10.5% 12.8% 12.8% 14.5% 13.7% 13.2%NTMV Cost ($) 737 962 1,175 1,601 2,311 5,251(NTMV Cost)/Cost 4.7% 4.8% 4.8% 5.6% 6.6% 8.6%MSRP NTMV 912 1,591 1,936 2,536 2,354 2,208Number of cars 41 49 36 46 28 33

percentage of N , the customer gets less. Let us denote this effect asthe dealer-strategy effect.

4. Data analysis

Some interesting observations can be made by comparing theaverages of various variables by MSRP group, reported in Table 2 .5

We can see that the negotiable range (which is also the hypo-

thetical markup) increases with the MSRP, and the ratio of thishypothetical markup to the cars cost is around 13%, being par-ticularly low (10.5%) for the lowest-price cars and particularly high(14.5%) for cars with MSRP of $30,000$35,000. The actualmarkup,NTMV Cost, also increases with the cars price. The ratio of theactual markup to the car s cost is about the same (around 4.8%)for cars in the three categories with MSRP up to $30,000, and thenit starts to increase with the car s MSRP, up to 8.6% in the groupof cars with MSRP above $45,000. That is, not only the price onwhich this markup percentage is taken increases, but also the per-centage itself increases, leading to large gains to the dealers fromselling these expensive cars. For example, cars with MSRP above$45,000 give the dealer on average an actual markup of $5,251 about 4.5 times the markup on a car with MSRP between $25,000

and $30,000, while the ratio of the average MSRP in the two groupsis only 2.4. 6

Finally, it is interesting to explore the two hypotheses formu-lated in the previous section. To do so, twosimple regressions wereused, where is the dependent variable. The difference betweenthe two regressions is whether the MSRP or the NTMV serves asthe cars price. Table 3 presents the results of the regressions. The

5 The range of prices that determines the groups was chosen by the followingalgorithm: startingwith the cheapestcar, each$5000 incrementwas examinedandwas chosenas therange if atleast20 cars fall withinthe range; if fewer than 20carsexisted in the range, the category was expanded by increasing its upper bound by$5000. Forexample, there areonly5 cars with MSRPbetween $10,000 and$15,000,so the range was expanded to $10,000$20,000 in order to include at least 20 cars

in the category.6 Thissomewhat resembles tippingpractices in restaurants.In theUS, thenorm in

upscalerestaurants isto tip20% ofthe bill,whilein family-stylerestaurants thenormis around 15% ( Post, 1997; Azar, 2004 ). This means that not only the bill is higherin upscale restaurants, resulting in higher dollar tips, but even the percentage tipsare higher in upscale restaurants. A possible reason for this is that the customersof upscale restaurants are much wealthier than customers of family-style restau-rants, and therefore they can afford to pay much more in tips even more thanproportionally to the meals cost ( Azar, 2004 ). Here, customers who can purchaseexpensive cars might be wealthier than those who purchase low-price cars, to theextent that the former group is not willing to spend much time comparing pricesin different dealerships or bargaining. In addition, wealthier customers might havestronger preferences for a specic model they want andbe less willing to settle foranother similar car (i.e., they view other alternative models as less substitutablefor their preferred model than customers who purchase low-priced cars). Conse-quently, the price competition between different dealers is much less intense forthehigher-price cars, leading to theresult that notonly thedollar markup is higher

formore expensivecars, but even thepercentage markup is higher.

Table 3Variables explaining success rate of customers in the bargaining process.

Dependent variable:

Constant 0.7218938(23.13)

0.731835(23.885)

Negotiable range 0.0000766(4.52)

0.0000647(3.24)

NTMV 0.0000151( 7.30)

MSRP 0.0000132( 5.86)

R2 0.309 0.228Observations 233 233

t -statistics obtainedwith robust standarderrorsare reportedin parentheses.All thecoefcients reported in thetableare statistically signicant at the1% level.

regressions reveal that in both specications the two hypothesesare strongly supportedby the data, allthe results being statisticallysignicant at p-values of 0.001 or less.

First, is negatively correlated with price, whether we measureprice by MSRP or NTMV. This gives support to the hypothesis aboutthe time-value effect: people who purchase more expensive carshave on average a higher time value and therefore they dedicateless time to bargaining, which results in receiving a smaller portionof the negotiable range (as measured by ). The coefcient beingabout 0.000014 ( 0.0000132 in one regression and 0.0000151in theother)meansthata $10,000 increasein thecars price reduces by 0.14, which implies that the customer obtains 14% less of the negotiable range (increasing the dealers prot by the sameamount). This is clearly an economically signicant magnitude.

Second, we can see that is positively correlated with thenegotiable range. 7 This is consistent with the second hypothe-sis suggested in the previous section, and can be a result of thenegotiable-range effect, of the dealer-strategy effect, or both. Thecoefcient being about 0.00007 (0.0000647 in one regression and0.0000766 in the other) means that a $1000 increase in the nego-tiable range increases by 0.07, which implies that the customerobtains 7% more of the negotiable range. This is certainly an eco-

nomically signicant magnitude.

5. Conclusion

Data about 233 different new car models were collected, anda measure of customer success in bargaining for a new car( ) was created by computing the ratio between the discountreceived on the MSRP and the total negotiable range (MSRP thedealers cost of the car). Two hypotheses were formulated about

7 Notice that if the amount of money customers could get as a discount on theMSRP was independent of the negotiable range, we should have obtained not justa zero correlation but a negative correlation between and the negotiable range,

because

isobtainedby dividing thediscount onthe MSRPby thenegotiablerange.

8/13/2019 Proiect Masini Articol

5/5

30 O.H. Azar / The Journal of Socio-Economics 41 (2012) 2630

bargaining behavior and outcomes in the new car market. The rsthypothesis was that customers who purchase more expensive carshave on average higherincome, thereforetheirtimevalueis higher,and therefore they would choose to stop bargaining sooner andwill succeed less in the bargaining process. This hypothesis wassupported by the data, that revealed a statistically signicant neg-ative correlation between and the MSRP or the NTMV (the actualaverage transaction price of the car).

The second hypothesis was that a positive correlation betweenthe negotiation range and should exist. This could result eitherfrom the customer having more incentives to negotiate when thenegotiable range is higher, or from thedealers insistence on a min-imal prot per car sold, where the elasticity of this minimal protwith respect to the negotiable range is less than 1. The data revealindeed a statistically signicant positive correlation between andthe negotiable range, in support of the second hypothesis.

References

Ayres, I.,Siegelman, P.,1995.Race andgenderdiscrimination in bargainingfor a newcar. American Economic Review 85, 304321.

Azar, O.H., 2004. What sustains social norms and how they evolve? The case of tipping. Journal of Economic Behavior and Organization 54, 4964.

Bragg, W.J., 2004. Car Buyers and Leasers Negotiating Bible, 3rd ed. Random HouseReference.

Busse, M., Silva-Risso, J., Zettelmeyer, F., 2006. $1000 cash-back: the pass-throughof auto manufacturer promotions. American Economic Review 96, 12531270.

Camerer,C., Babcock, L., Loewenstein,G., Thaler, R., 1997. Labor supply of New YorkCity cabdrivers: one day at a time. The Quarterly Journal of Economics 112,407441.

Ford, B., 2005. What Car Dealers Wont Tell You. Plume.Goff, B.G., Boles, J.S., Bellenger, D.N., Stojack, C., 1997. The inuence of salesperson

selling behaviors on customer satisfaction with products. Journal of Retailing73, 171183.

Gneezy, U., Haruvy, E., Roth, A., 2003. Bargaining under a deadline: evidence fromthe reverse ultimatum game. Games and Economic Behavior 45, 347368.

Goldberg, P.K., 1996. Dealer price discrimination in new car purchases: evidencefrom the Consumer Expenditure Survey. Journal of Political Economy 104,622654.

List,J.A.,2004. Thenatureand extent of discriminationin themarketplace:evidencefrom theeld. Quarterly Journal of Economics 119, 4989.

Lyle, P.,1999. Howto Buyor Lease a Car Without Getting Ripped Off. Adams Media.Moran, S.,Ritov,I., 2002. Initial perceptionsin negotiations:evaluationand response

to logrolling offers. Journal of Behavioral Decision Making 15, 101124.Post, P., 1997. Emily Posts Etiquette, 16th ed. HarperCollins, New York.Reed, P.,2005.Edmunds.comStrategiesfor Smart CarBuyers. EdmundsPublications.Rubinstein, A., 1982. Perfect equilibrium in a bargaining model. Econometrica 50,

97110.Verhoef,P.C., Langerak, F., Donkers,B., 2007. Understandingbrandand dealer reten-

tion in thenew carmarket:the moderatingroleof brand tier.Journal of Retailing83, 97113.