Cite as: S. Zinn; 2012; A mate-matching algorithm for continuous-time microsimulation models; International Journal of Microsimulation; 5(1); 31–51. doi: 10.34196/ijm.00066
Abstract
The timing of partnership formation in closed continuous-time microsimulation models poses difficulties due to the continuous time scale. in this paper the problem is resolved by the concept of a partnership market, which individuals can enter and leave at any point in time over the complete simulation time range. Each individual, who looks for a spouse, remains in the market for a specific period during which searching and matching is performed. To build up synthetic couples, the model imitates a decision making-process. The decision to enter a partnership depends on empirically estimated logit models for the probability that a given woman and a given man will get together, and also on an individual aspiration level regarding a potential partner. A couple is formed if a positive decision has been made and the timing of the partnership formation is consistent with the individual searching periods of the prospective spouses. The algorithm is illustrated by an example in which simulations are run to project a synthetic population, similar to the population of the Netherlands, by using an extended version of the microsimulation tool of the MicMac project.
This article contains 10 figures and 3 tables.
Figures
Figure 1. Classification of mating models and mate-matching algorithms for microsimulation.Figure 2. Woman I1 experiences a marriage event at time t1. Man I2 experiences a marriage event at time t2. As t1 ∈ [t2 − 0.5 years, t2 + 0.5 years] and t2 ∈ [t1 − 0.5 years, t1 + 0.5 years] both individuals have overlapping searching periods and might meet during the mating process.Figure 3. Densities of the beta distributions that are used to determine aspiration levels regarding partners. The densities vary with gender and age. For females, four different curves are applied: one below age 18, one for ages between 18 and 30, one between ages 30 and 35, and one after age 40. For males, two different curves are applied: one for males younger than 30 and the other after age 30.Figure 4. Re-estimation of transition rates of childless females with a lower secondary (medium) education who experience a transition from “being single after leaving parental home” (nSl) to “first cohabitation” (nCO).Figure 5. Re-estimation of transition rates of highly educated males who experience a transition from “being single after leaving parental home” (nSI) to “married” (nMA).Figure 6. Age distribution of unsuccessful seekers at the time when they enter the partnership market.Figure 7. Number of unsuccessful seekers according to the year when they initialize a partner search.Figure 8. Differences in the educational level of spouses in observed and simulated couples. Each bar shows the percentage of females in the corresponding category.Figure 9. Age differences of spouses in observed and simulated couples.Figure 10. Comparison of re-estimated transition rates of highly educated males who experience a transition from “being single after leaving parental home” (nSI) to “married” (nMA). The left graph shows the empirical input rates used. The graph in the middle displays re-estimated rates from a simulation run with B = 2 years (intersection of the searching periods), and the right graph shows re-estimated rates from a simulation run with B = 0.25 years.
Tables
Table 1. Parameters and suggested parameter values for the present stochastic mate-matching procedure.
