Abstract
While there is a great deal of theoretical and experimental literature on what factors affect bargaining outcomes, there is little empirical work based on data from real markets. In this paper we analyze negotiations for new cars, a $340 billion industry in the United States in 2010. Our results suggest that search costs, incomplete information, and bargaining disutility have an economically significant effect in real-world negotiations: we estimate that relative to an uninformed consumer, a consumer with basic information about the seller’s reservation price and his own outside options captures 15% of the average dealer margin from selling an automobile. We also find that a buyer’s search cost and bargaining disutility have significant effects on bargaining outcomes. Finally, our results show that while search is common, there remains a substantial group of consumers who do not engage in any of the search behaviors we measure. We hypothesize that these buyers are not aware of how easy and effective certain activities in improving negotiation outcomes can be.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
There is, or course, a large non-experimental empirical literature in economics on labor negotiations between unions and firms (Kennan and Wilson 1993).
The difference is statistically significant, as are all of the findings presented in this introduction.
The models in this subsection don’t consider search for outside options or different levels of bargaining disutility—we discuss these in the next subsections.
While there are papers which consider two-sided incomplete information in a dynamic model, their concern is with efficiency and not how varying degrees of incomplete information affects the division of surplus (see, for example, Ausubel and Deneckere 1998).
Dealers provide their data to DSA in exchange for information about local market conditions.
The error term in the price equation might nonetheless absorb factors important to a negotiation that we do not measure, such as whether the consumer has a friendly smile or is attractive. We have no reason to think, however, that a “winning smile” and similar consumer characteristics are correlated with the search behaviors of consumers. The only reason one might think that a friendly smile would affect search is a version of the Peltzman seat-belt argument (Peltzman 1975): If a buyer knows that her smile will lower the price of the car, she might exert somewhat less effort in searching than she otherwise would. Such correlation would result in OLS estimates being biased downwards and would thus be biased against our findings.
Dealer-installed accessories that contribute to the resale value include items such as upgraded tires or a sound system, but would exclude options such as undercoating or waxing.
All dealers in California are charged the same invoice price by the manufacturer.
“Holdback” is the industry term for a percentage of the invoice price that is held by the manufacturer for a period and then rebated to the dealer. It serves the purpose of creating a small margin for the dealer even if he sells the car at the invoice price.
We wanted to pre-test the survey with 250 consumers of one car model. Among cars that sold approximately 250 units in our transaction data, the Jeep Grand Cherokee fell closest to the average purchase price of new cars.
This does not exclude that any specific dealer might have low inventory of a particular model on a particular day; any price response to random inventory shocks will be absorbed in the error term.
We asked about the number of dealers of the same nameplate because we knew from prior work that price competition occurs primarily between dealers of the same nameplate. We also collected information on the total number of dealers visited, irrespective of their nameplate. The results of the paper do not change substantially with this alternative measure. This is not surprising given that the average response category for total dealers visited in our sample is 2.97, while the average response category for visited dealers of the focal brand is 2.61. Response categories: 1 (0 dealers), 2 (1 dealer), 3 (2–3 dealers), 4 (4–6 dealers), and 5 (7 or more dealers).
One might be concerned that buyers who indicate apprehension about the bargaining process do so because they happen to have negotiated a bad price, not-—as we will suggest—the other way around. In the empirical analysis on page 29 we will explicitly control for this possibility with a variable that measures independently how well the buyer thought he or she did in the negotiation.
The rotated factor loadings (varimax rotation) for the first factor are 0.82 for DoPriceComparisons, 0.72 for InternetForInfo, and 0.85 for GatherMuchInfo. For the second factor, corresponding to “car knowledge,” the factor loadings are 0.85 for ReadCarMagazine, and 0.84 for VisitDealerForFun. For the third factor, corresponding to “bargaining disutility,” the factor loadings are 0.73 for AfraidTakenAdvantage, and 0.78 for NoTimeToShop.
Note that these sources are not mutually exclusive, so the percentages add up to more than 100.
In this and other specifications we did not any significant effect of our “Competition” variable. We think this due to our small sample size. Using a similar but much larger dataset (650,950 observations), Scott Morton et al. (2003) find that one additional dealer of the same nameplate within a 10 mile radium is associated with approximately 0.2% lower prices.
Using a larger sample of (similar) transaction data we found in Scott Morton et al. (2003) that not just gender but also race were significant predictors of price. Specifically, we found a 0.8% price premium for blacks and a 0.6% price premium for Hispanics. Looking across all specifications in Table 3 reveals that the point estimates of “Black” and “Hispanic” are not inconsistent with these prior results. However, the vastly smaller sample size (1,402 vs. 650,850 observations) in this study does not allow us to reject the hypothesis that the coefficients are zero. Turning to gender, the female coefficient in Scott Morton et al. (2003) was 0.2%. This is also not inconsistent with the findings in Table 3, although no specification in column 2 allows us to conclude that the female coefficient is different from zero at a 5% level. Finally, Scott Morton et al. (2003) found that age and income were also significant predictors of price. In the smaller sample considered here, the coefficient on income is negative but statistically insignificant (− 0.27, p-value 0.30). The coefficient on squared income (0.04, p-value 0.055) implies that those with the highest incomes pay more, all else equal.
We can further control for demographic differences by allowing for zip-code fixed effects. While these are not at the level of the individual, they are likely to control for unobserved consumer characteristics that are not measured in our demographic variables. We find that including 641 zip-code fixed effects somewhat increase the magnitude of the “Informed” coefficient relative to the estimate in column 2. Specifically, the coefficient on ‘Informed’ goes from −1.14 to −1.42 and remains significant (p-value < 0.01). Because zip-code level fixed effects use up so many degrees of freedom we don’t include them in the remaining specifications.
The consumer traits we construct are based in part on consumers’ assessment of their bargaining disutility. These assessments were made six to twelve weeks after the consumers purchased a car. If consumers infer their bargaining disutility from the price they obtained for this particular vehicle, there could be an endogeneity between prices and consumer traits. Please see page 29 for an investigation of this issue.
It is perhaps surprising that buyers who claim to have learned the “fair price or market value” do not do as well as buyers who say that they learned the invoice price. We conjecture that this is because the term “fair price” is much more open to interpretation by respondents. If “fair market price” is included without the other three types of information, it is negative and significant at the 3% level, suggesting it also is collinear with invoice price.
The dealer fixed effects are jointly significant, F(170, 1,199) = 36.6, p-value < 0.01.
Alternatively, we can also sort consumers by constructing a variable that is the sum of the normalized values of the responses to the three measures of consumers’ willingness to search. For each variable we calculate the mean and standard deviation over all respondents. Then we normalize the answer for each individual by subtracting the mean and dividing by the standard deviation.
Notice that while our data allows us to infer that consumer actions (e.g., getting to know invoice price, going to additional dealers, or having a competing offer) yield the outcomes predicted by the bargaining theory literature, our data cannot conclusively answer whether the mechanisms that links these actions to better bargaining outcomes are the same mechanisms that are modeled in the bargaining theoretic papers we discuss. For example, the models in the first part of Section 2 imply that knowing the invoice price of the dealer will help the consumer in the negotiation because it reduces uncertainty about the reservation price of the dealer. However, it could be that knowing the invoice price helps consumers because it signals to the dealer that the consumer searches a lot and thus perhaps will visit competing dealers (we thank an anonymous referee for pointing this out). We cannot conclusively distinguish between these different mechanisms because—while we have data from a consumer survey—we do not have corresponding data from a dealer survey. Nonetheless, we can present some evidence that suggests that our results may not be driven by signaling. We can do this because we know from the survey which consumers told dealers about their search and dealer visits. When we exclude from the sample consumers who stated that they told the dealer about obtaining an invoice price, and/or obtaining a competing offer, we obtain results that are similar to our main findings. Details are available from the authors. We recognize that dealers may still make inferences even if consumers did not explicitly tell them about having searched. However, we think that the signaling issue is less likely to be present in the restricted sample.
We cannot measure and thus abstract from the gains consumers obtain from searching to find a car that matches their needs.
Only 37% of buyers who reported not having a high school degree used the Internet. This is in contrast to 81% of buyers with a college degree or higher. Eighty-seven percent of buyers with income above $150,000 but only 47% of buyers with income between $20,000 and $29,999 reported using the Internet for car buying.
To the extent the salesperson offers an obviously “too high” price and the consumer gets disutility from pointing this out, there may be a disutility associated with collecting the offer.
Consumers who did not visit an additional dealership selling the car model they ended up buying are much less likely to have a competing offer (22 versus 54%). This is consistent with a standard value of time explanation: it is costly for them to generate a competing offer without a second dealer visit.
For laboratory-based experimental papers that discuss how the information available to the bargaining parties affects bargaining outcomes, we refer the reader to Valley et al. (1992) and Croson et al. (2003). For an experimental paper that analyzes the role of search for information about outside options in a bargaining setting, please see Zwick and Lee (1999). If we interpret a higher bargaining disutility as a higher per-period bargaining cost, Rapoport et al. (1990) and Weg and Zwick (1991) present experiments about the effect of bargaining disutility on bargaining outcomes. For a detailed discussion of the experimental literature please see Thompson (1990), Kennan and Wilson (1993), and Roth (1995).
Please see Scott Morton et al. (2003) for a discussion of how the various race and gender results in the literature compare.
To understand why this is necessary, suppose wealthy individuals are more likely to buy houses with central air-conditioning, an attribute that is unobservable to the researcher. If such houses sell for a premium compared to houses without central air-conditioning, the researcher will concluded that wealthy individuals are worse off in price negotiations than other individuals, controlling for all observable product characteristics.
References
Ausubel, L. M., Cramton, P., & Deneckere, R. J. (2002). Bargaining with incomplete information. In R. J. Aumann, & S. Hart (Eds.), Handbook of game theory (Vol. 3). Amsterdam: Elsevier.
Ausubel, L. M., & Deneckere, R. J. (1998). Bargaining and forward induction. mimeo, University of Maryland University of Wisconsin-Madison.
Ayres, I., & Siegelman, P. (1995). Race and gender discrimination in bargaining for a new car. American Economic Review, 85(3), 304–321.
Beatty, S. E., & Smith, S. M. (1987). External search effort: An investigation across several product categories. Journal of Consumer Research, 14(1), 83–95.
Busse, M., Silva-Risso, J., & Zettelmeyer, F. (2006). $1000 cash bank: The pass- through of auto manufacturer promotions. American Economic Review, 96(4).
Chatterjee, K., & Lee, C. C. (1998). Bargaining and search with incomplete information about outside options. Games and Economic Behavior, 22, 203–237.
Chatterjee, K., & Samuelson, L. (1983). Bargaining under incomplete information. Operations Research, 31(5), 835–851.
Chen, Y., Yang, S., & Zhao, Y. (2003). A structural approach to modeling negotiated prices of automobiles. New York: New York University.
Chikte, S. D., & Deshmukh, S. D. (1987). The role of external search in bilateral bargaining. Operations Research, 35(2), 198–205.
Cramton, P. (1985). Sequential bargaining mechanisms. In A. Roth (Ed.), Game theoretic models of bargaining. Cambridge: Cambridge University Press.
Croson, R., Boles, T., & Murnighan, J. K. (2003). Cheap talk in bargaining experiments: Lying and threats in ultimatum games. Journal of Economic Behavior and Organization, 51, 143–159.
Fudenberg, D., Levine, D. K., & Tirole, J. (1985). Infinite-horizon models of bargaining with one-sided incomplete information. In A. Roth (Ed.), Game theoretic models of bargaining. Cambridge: Cambridge University Press.
Fudenberg, D., & Tirole, J. (1991). Game theory. Cambridge: MIT Press.
Gul, F., Sonnenschein, H., & Wilson, R. (1986). Foundations of dynamic monopoly and the coase conjecture. Journal of Economic Theory, 39, 155–190.
Harding, J., Rosenthal, S., & Sirmans, C. (2003). Estimating bargaining power in the market for existing homes. Review of Economics and Statistics, 85(1), 178–188.
Keith, K., & McWilliams, A. (1999). The returns to mobility and job search by gender. Industrial and Labor Relations Review, 52(3), 460–477.
Kennan, J., & Wilson, R. (1993). Bargaining with private information. Journal of Economic Literature, 31(1), 45–104.
Kiefer, N., & Neumann, G. R. (1979). An empirical job-search model, with a test of the constant reservation-wage hypothesis. Journal of Political Economy, 87(1), 89–107.
Lee, C. C. (1994). Bargaining and search with recall: A two-period model with complete information. Operations Research, 42(6), 1100–1109.
Muthoo, A. (1995). On the strategic role of outside options in bilateral bargaining. Operations Research, 43, 292–297.
Narendranatha, W., & Nickell, S. (1985). Modeling the process of job search. Journal of Econometrics, 28(1), 29–49.
Peltzman, S. (1975). The effects of automobile safety regulation. Journal of Political Economy, 83(4), 677–726.
Perry, M. (1986). An example of price formation in bilateral situations: A bargaining model with incomplete information. Econometrica, 54(2), 313–321.
Punj, G., & Staelin, R. (1983). A model of consumer information search behavior for new automobiles. Journal of Consumer Research, 9, 366–80.
Rapoport, A., Weg, E., & Felsenthal, D. S. (1990). The effects of fixed costs in two-person sequential bargaining. Theory and Decision, 28, 47–72.
Ratchford, B., & Srinivasan, N. (1993). An empirical investigation of returns to search. Marketing Science, 12(1), 73–87.
Ratchford, B. T., Talukdar, D., & Lee, M.-S. (2003). The impact of the Internet on information search for automobiles. Journal of Marketing Research, 40(2), 193–209.
Roth, A. E. (1995). Bargaining experiments. In J. H. Kagel, & A. E. Roth (Eds.), The handbook of experimental economics. Princeton: Princeton University Press.
Rubinstein, A. (1982). Perfect equilibrium in a bargaining model. Econometrica, 50, 97–109.
Scott Morton, F., Zettelmeyer, F., & Silva-Risso, J. (2003). Consumer information and discrimination: Does the Internet affect the pricing of new cars to women and minorities?, Quantitative Marketing and Economics, 1(1), 65–92.
Srinivasan, N., & Ratchford, B. T. (1991). An empirical test of a model of external search for automobiles. Journal of Consumer Research, 18(2), 233–42.
Stern, S. (1989). Estimating a simultaneous search model. Journal of Labor Economics, 7(3), 348–369.
Stigler, G. J. (1961). The economics of information. Journal of Political Economy, 69(3), 213–225.
Thompson, L. (1990). Negotiation behavior and outcomes: Empirical evidence and theoretical issues. Psychological Bulletin, 108(3), 515–532.
Valley, K., Blount White, S., Neal, M., & Bazerman, M. (1992). Agents as information brokers: The effects of information disclosure on negotiated outcomes. Organizational Behavior and Human Decision Processes, 51, 220–236.
Weg, E., & Zwick, R. (1991). On the robustness of perfect equilibrium in fixed cost sequential bargaining under an isomorphic transformation. Economics Letters, 36(1), 21–24.
Wolinsky, A. (1987). Matching, search, and bargaining. Journal of Economic Theory, 42, 311–333.
Zettelmeyer, F., Scott Morton, F., & Silva-Risso, J. (2006). How the internet lowers prices: Evidence from matched survey and auto transaction data. Journal of Marketing Research, 43(2), 168–181.
Zwick, R., & Lee, C. C. (1999). Bargaining and search: An experimental study. Group Decision and Negotiation, 8, 463–487.
Zwick, R., Rapoport, A., Lo, A., & Muthukrishnan, A. (2003). Consumer sequential search: Not enough or too much? Marketing Science, 22(4), 503–519.
Author information
Authors and Affiliations
Corresponding author
Electronic Supplementary Material
Below is the link to the electronic supplementary material.
Rights and permissions
About this article
Cite this article
Scott Morton, F., Silva-Risso, J. & Zettelmeyer, F. What matters in a price negotiation: Evidence from the U.S. auto retailing industry. Quant Mark Econ 9, 365–402 (2011). https://doi.org/10.1007/s11129-011-9108-1
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11129-011-9108-1