Abstract
We generalize the classic dynamic single-item search model to a setting with multiple items and vector offers for subsets of items. We first show a computationally feasible way to solve the dynamic optimization problem, and then prove structural results. Although assignment is not generally monotonically increasing in offer value, we show that, in a special case “additive” model, monotonicity holds if costs are submodular. We examine how the thresholds for assignment change with the remaining items, and whether there are gains to grouping searches. Finally, we consider a stopping rule version of the problem with no subsets for sale, showing the optimal policy is myopic.
Similar content being viewed by others
Notes
Precisely, f is a submodular set function if, for sets S and T, \(f(S \cup T) + f(S \cap T) \le f(S) + f(T)\), and f is supermodular if \(f(S \cup T) + f(S \cap T) \ge f(S) + f(T)\).
Bruss and Ferguson (1997) also do not provide our solution to the optimality equation or computational approach. In addition, we extend their results about the stopping rule variation of the additive model. While Bruss and Ferguson (1997) show that a one-stage lookahead policy is optimal for the stopping rule variation of the additive model, we show that it is also optimal in the single applicant model, which is not a simple extension as the intuition is entirely different.
Specifically, the positions (items) should be interpreted here as workers and the applicants as sequentially arriving jobs, with the values of the value vector \(X_i\) being the value of assigning the job with vector \((X_1, \ldots ,X_n)\) to worker i.
Specifically, DLR supposed that there are n workers, with worker i having a specified value \(p_i\). Jobs appear sequentially, with each job having a value X that is chosen, independently from job to job, according to a known distribution function. If a worker with value \(p_i\) is assigned to a job with value x then a return \(p_i x\) is earned and the problem was to assign workers to jobs so as to maximize the expected sum of values. (DLR assumed that each job had to be assigned to some worker so the problem ended when n jobs had appeared.) Thus the model of DLR is a special case of our model in which the relative value of the applicants is fixed across periods, i.e., random value vectors are of the special form \((p_1X, p_2X, \ldots , p_n X)\) with X having a specified distribution.
Note that the additive model does allow the firm to take into account any complementarities between applicants and existing employees; these complementarities would be reflected in each applicant’s match value.
Note that even the more general model does not allow for complementarities across applicants who apply in different periods; this would require that the value of the value vector change as a result of past hiring decisions which would fundamentally change the nature of the classic search model.
References
Arnold, M., & Lippman, S. (1995). Selecting a selling institution: auctions versus sequential search. Economic Inquiry, 33(1), 1–23.
Bergemann, D., & Välimäki, J. (2010). The dynamic pivot mechanism. Econometrica, 78(2), 771–789.
Board, S., & Skrzypacz, A. (2010). Revenue management with forward-looking buyers. Stanford: Stanford University.
Boshuizen, F., & Gouweleeuw, J. (1993). General optimal stopping theorems for semi-Markov processes. Advances in Applied Probability, 25, 825–846.
Bruss, F., & Ferguson, T. (1997). Multiple buying or selling with vector offers. Journal of Applied Probability, 34, 959–973.
David, I., & Levi, O. (2001). Asset-selling problems with holding costs. International Journal of Production Economics, 71(1), 317–321.
Derman, C., Lieberman, G., & Ross, S. (1972). A sequential stochastic assignment problem. Management Science, 18(7), 349–355.
Derman, C., & Sacks, J. (1960). Replacement of periodically inspected equipment. (an optimal optional stopping rule). Naval Research Logistics Quarterly, 7(4), 597–607.
Gallego, G., & Van Ryzin, G. (1994). Optimal dynamic pricing of inventories with stochastic demand over finite horizons. Management Science, 40(8), 999–1020.
Gallien, J. (2006). Dynamic mechanism design for online commerce. Operations Research, 54(2), 291–310.
Gershkov, A., & Moldovanu, B. (2009). Dynamic revenue maximization with heterogeneous objects: A mechanism design approach. American Economic Journal: Microeconomics, 1(2), 168–198.
Iwata, S. (2002). A fully combinatorial algorithm for submodular function minimization. In Proceedings of the thirteenth annual ACM-SIAM symposium on discrete algorithms (pp. 915–919). Society for Industrial and Applied Mathematics.
Lippman, S., & McCall, J. (1976). The economics of job search: A survey. Economic Inquiry, 14(2), 155–189.
MacQueen, J., & Miller, R, Jr. (1960). Optimal persistence policies. Operations Research, 8(3), 362–380.
McMillan, J., & Rothschild, M. (1994). Search (Chapter 27). In R. Aumann & S. Hart (Eds.), Handbook of game theory with economic applications (Vol. 2). New York: North Holland.
Ross, S. (1983). Introduction to stochastic dynamic programming: Probability and mathematical. London: Academic Press, Inc.
Roth, A. E., & Sotomayor, M. A. O. (1992). Two-sided matching: A study in game-theoretic modeling and analysis (Vol. 18). Cambridge: Cambridge University Press.
Shapley, L. (2006). Complements and substitutes in the optimal assignment problem. Naval Research Logistics Quarterly, 9(1), 45–48.
Sofronov, G. (2013). An optimal sequential procedure for a multiple selling problem with independent observations. European Journal of Operational Research, 225(2), 332–336.
Stigler, G. (1961). The economics of information. The Journal of Political Economy, 69, 213–225.
Stigler, G. (1962). Information in the labor market. The Journal of Political Economy, 70, 94–105.
Toroslu, I., & Üçoluk, G. (2007). Incremental assignment problem. Information Sciences, 177(6), 1523–1529.
Zuckerman, D. (1986). Optimal stopping in a continuous search model. Journal of Applied Probability, 23, 514–518.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We appreciate support from the National Science Foundation (Ross through CMMI-1662442) and from the endowment in memory of B.F. Haley and E.S. Shaw (Dizon-Ross)
Rights and permissions
About this article
Cite this article
Dizon-Ross, R., Ross, S.M. Dynamic search models with multiple items. Ann Oper Res 288, 223–245 (2020). https://doi.org/10.1007/s10479-019-03472-z
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10479-019-03472-z