Abstract
We discuss Herzberg’s (Theory Decis 78(2):319–337, 2015) treatment of linear aggregation for profiles of infinitely many finitely additive probabilities and suggest a natural alternative to his definition of linear continuous aggregation functions. We then prove generalizations of well-known characterization results due to (J Am Stat Assoc 76(374):410–414, 1981). We also characterize linear aggregation of probabilities in terms of a Pareto condition, de Finetti’s notion of coherence, and convexity.
Similar content being viewed by others
Notes
Recall that if X is a set and \({\mathcal {X}}\) is an algebra of subsets of X, then to say that a probability \(P: {\mathcal {X}} \rightarrow [0,1]\) is finitely additive means that \(P(\cup _{i=1}^n A_i) = \sum _{i=1}^n P(A_i)\) for all \(n \in {\mathbb {N}}\) and pairwise disjoint \(A_1,\ldots ,A_n \in {\mathcal {X}}\). If, in addition, \(P(\cup _{i=1}^\infty A_i) = \sum _{i=1}^\infty P(A_i)\) whenever \(A_1, A_2,\ldots \in {\mathcal {X}}\) are pairwise disjoint and \(\cup _{i=1}^\infty A_i \in {\mathcal {X}}\), then P is countably additive.
Herzberg adopts McConway’s framework, according to which aggregation functions are defined for all algebras over \(\Omega \) as opposed to a fixed algebra \({\mathcal {A}}\). Although our definition may appear less general, it is clear that our results apply in the McConway–Herzberg framework by systematically quantifying over algebras of \(\Omega \). The framework with a fixed algebra \({\mathcal {A}}\) is both standard (Genest and Zidek 1986; Stewart and Ojea Quintana 2018) and notationally simpler.
Our presentation follows Stewart and Ojea Quintana (2018).
Note that [Q] is nonempty and need not be a singleton. See Rao and Rao (1983, Chapter 3).
See Aliprantis and Border (2006, 11.2).
See, for example, Schirokauer and Kadane (2007).
Wagner (1982, Theorem 7) discovered the equivalence of (b) and (c) independently of McConway using essentially the same proof technique.
To see this, it may help to observe that, by the linearity of the integral, F is Pareto if and only if for all step functions s, \(\int _\Omega s {\text {d}}P_i \ge 0\) for all \(i \in I\) implies \(\int _\Omega s {\text {d}}F(\underline{P}) \ge 0\). A detailed study of results like Proposition 1 can be found in Cassese (2018). Related results appear in Dubins (1975) and Lane and Sudderth (1984).
See Berti et al. (2013, p. 50) and references therein.
A referee points out that connections between coherence and Harsyani’s utilitarian aggregation theorem have been explored in Diecidue and Wakker (2002) and Diecidue (2006). These papers differ from the present one in two important respects. First, their results are stated for finite I, whereas I is arbitrary in Definition 8. Second, our notion of coherent aggregation is more closely analogous to coherence in de Finetti’s sense.
\({\mathcal {D}}\) denotes a directed set. See Aliprantis and Border (2006, 2.4) for an overview of nets.
If \(x \in X\), then \(\delta _x \in \Delta (2^X)\) denotes point mass at x, which is defined by \(\delta _x(A) = 1_A(x)\) for all \(A \subseteq X\).
References
Aczél, J. (2006). Lectures on functional equations and their applications. North Chelmsford: Courier Corporation.
Aliprantis, C. D., & Border, K. C. (2006). Infinite dimensional analysis: A Hitchhiker’s guide. Berlin: Springer.
Berti, P., Pratelli, L., & Rigo, P. (2013). Finitely additive equivalent martingale measures. Journal of Theoretical Probability, 26(1), 46–57.
Berti, P., Pratelli, L., & Rigo, P. (2014). A unifying view on some problems in probability and statistics. Statistical Methods & Applications, 23(4), 483–500.
Cassese, G. (2017). The representation of conglomerative functionals. In M. B. Ferraro, P. Giordani, B. Vantaggi, M. Gagolewski, M. Á. Gil, P. Grzegorzewski, & O. Hryniewicz (Eds.), Soft methods for data science (Vol. 456, pp. 103–111). Berlin: Springer.
Cassese, G. (2018). Conglomerability and representations. Journal of Convex Analysis, 25(3), 789–815.
de Finetti, B. (1974). Theory of Probability, (Vol. 1). Hoboken: Wiley.
Diecidue, E. (2006). Deriving Harsanyi’s utilitarianism from de Finetti’s book-making argument. Theory and Decision, 61(4), 363–371.
Diecidue, E., & Wakker, P. P. (2002). Dutch books: Avoiding strategic and dynamic complications, and a comonotonic extension. Mathematical Social Sciences, 43(2), 135–149.
Dubins, L. E. (1975). Finitely additive conditional probabilities, conglomerability and disintegrations. The Annals of Probability, 3(1), 89–99.
Genest, C., & Zidek, J. V. (1986). Combining probability distributions: A critique and an annotated bibliography. Statistical Science, 1(1), 114–135.
Harsanyi, J. C. (1955). Cardinal welfare, individualistic ethics, and interpersonal comparisons of utility. Journal of Political Economy, 63(4), 309–321.
Herzberg, F. (2015). Aggregating infinitely many probability measures. Theory and Decision, 78(2), 319–337.
Lane, D. A., & Sudderth, W. D. (1984). Coherent predictive inference. Sankhyā: The Indian Journal of Statistics. Series A: The Indian Journal of Statistics, Series A, 46(2), 166–185.
McConway, K. J. (1981). Marginalization and linear opinion pools. Journal of the American Statistical Association, 76(374), 410–414.
Oxtoby, J. C. (2013). Measure and category: A survey of the analogies between topological and measure spaces (2nd ed.). Berlin: Springer.
Rao, K . B., & Rao, M . B. (1983). Theory of charges: A study of finitely additive measures, volume 109 of pure and applied mathematics. Cambridge: Academic Press.
Schirokauer, O., & Kadane, J. B. (2007). Uniform distributions on the natural numbers. Journal of Theoretical Probability, 20(3), 429–441.
Stewart, R. T., & Ojea Quintana, I. (2018). Probabilistic opinion pooling with imprecise probabilities. Journal of Philosophical Logic, 47(1), 17–45.
Thill, M. (2008). Closed convex hulls, contents, and k-spectral states. Mathematische Nachrichten, 281(2), 297–303.
Wagner, C. (1982). Allocation, Lehrer models, and the consensus of probabilities. Theory and Decision, 14(2), 207–220.
Zhou, L. (1997). Harsanyi’s utilitarianism theorems: General societies. Journal of Economic Theory, 72(1), 198–207.
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.
Rights and permissions
About this article
Cite this article
Nielsen, M. On linear aggregation of infinitely many finitely additive probability measures. Theory Decis 86, 421–436 (2019). https://doi.org/10.1007/s11238-019-09690-y
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11238-019-09690-y