Slate has an article by Mark Gimein attempting to explain the shortage of available, appealing men in the marriage market. Tyler Cowen doesn't buy it, arguing that settling for less becomes psychologically harder the longer one holds out (also re-iterating his argument that people should "settle").

Slate's Gimein argues that we can view the marriage market as a bidding between two distinct distributions of females: strong bidders who think they have above-average intelligence, beauty, health, and wealth; and weak bidders who think they do not. As previously shown, weak bidders tend to outbid strong bidders; Gimein believes this previous result applies: given multiple "auctions", strong-bidding females will hold out while weak-bidding females essentially grab the first thing that comes along.

The primary flaw with this analysis lies in its reversibility. Gimein brazenly (other adverbs may apply ;) hints at this with:

Obviously, this is simplified -- in contemporary life, both sides get plenty of chances to be selective.

One can also view the marriage market as males bidding for females, with similar results. For any grouping of males over time, weak-bidding males should drop out of competition as they essentially grab the first thing that comes along, leaving incrementally rising concentrations of strong-bidding males.

Additionally, Gimein cites no data in his assertion of a failing marriage market. Googling for population "unmarried women" "unmarried men" yields (1st result) US Census 2000 data for the ratio of unmarried men per 100 unmarried women:

Never Married Ratio of unmarried men to women Male 32,381,377 86 Female 27,531,993

How do we get from a 5M bachelor surplus to a ratio of less than one? By counting Widowed and Divorced: Since females live ~9 years longer than males and the average age-difference has been steady at 2-3 years; and the apparently unhappy result that guys appear to have an easier time (or even want to) remarrying, we see that the data do not at first level analysis indicate a marriage market running short of guys.

Interestingly, the ratios of never-married men to never-married females vary drastically from state to state

State/Non-stateNever-Married Males to Never-Married Females
District of Columbia0.96
New York1.07
Rhode Island1.07
South Carolina1.13
New Jersey1.13
New Mexico1.16
New Hampshire1.18
North Carolina1.2
West Virginia1.23
South Dakota1.28
North Dakota1.36

Never-married also includes people who actually do not wish to marry, which poses a secondary problem as sexual preferences may sort geographically. Assuming Gimein's observation holds for the population in his area, he most likely lives in either a sexual-preference-skewed area and/or someplace like Alaska.

Let's try to model the marriage market given (as best I know) contemporary evolution and economics. Assume we have two groups of females and males (possibly differing in number) normally distributed along several dimensions: age, health, beauty, intelligence, demeanor, sexual preference, genetic difference, and financial assets.

Relative to each population, an individual with a significantly higher aggregate score may choose to pursue a strategy of repeated selective mating with different individuals to maximize the genetic diversity of their offspring (e.g. Wilt Chamberlain), while others seek to pair for mutual support (or blocking access depending on your viewpoint).

Loners and Pairers will have distinct mating strategies when they live someplace with strong legal backing for sharing the burden of child-bearing:

  1. Lone males with few seizable financial assets (states do not currently seize the other assets (beauty, intelligence, etc.)) will mate opportunistically as the state will fail to transfer non-existent assets from them.
  2. Lone males with seizable financial assets will seek to mate with as many high-value females as he can budget. Certain mutually-beneficial matings may occur with financially-secure females.
  3. Lone females with few financial assets will mate with high-scoring males with seizable financial assets as they have a much more limited number of times they mate with the intent of producing offspring (duration of fertility / (duration of pregnancy)).
  4. Lone females with secure financial standing have more options than females without financial assets as they can support offspring independently.

However, contrary to evolution, Loners may not choose mating strategies in order to maximize genetic offspring count and diversity. The breakdown of those variables currently stands as an open question in the field (as far as I know).

For Pairers, each individual should try to find the highest value individual they can attract (taking search costs into account). Each individual will try to signal their value along each of the attribute dimensions. Some of the attributes' signals will be more fakable than others (compare genetic diversity or health to either wealth or beauty).

Interestingly, from a possibly massively biased survey of friends last night, roughly half (of 7) felt that buying better signals increased odds, and roughly half felt that odds were not increased significantly. Lying about signal faking would be evolutionarily rewarded, hence some possible massive skew aside from the small non-random sample. However, should this exist on a larger scale in society, signal faking Pairers may attract non-faking others and thereby have a larger selection of mates.

On a side note, the deception may go further, given the information asymmetry involved historically in pregnancy (female knows the baby has her genes, meanwhile the Paired male doesn't). This asymmetry leads to cheating where the female mates with a higher-status male and the Paired male bears the burden of raising the child for at least as long as the child is in utero.

Of the attributes like genetic diversity and demeanor, we have a n-way pairings possible. Within the many facets of demeanor, people may assort similarly (happy-go-lucky to happy-go-lucky) or dissimilarly (passive to aggressive). Additionally, within genetic diversity individuals use pheromones (indicating protein variation) assorting with pheromones different than one's own (of the phenotypes we can visually appreciate like skin color, height, etc. the vast bulk of genetic diversity lies in non-visible items like which proteins our body uses and exudes).

The Pairer's selection process then has two parts: 1) of the n-way pairing candidates for genetic diversity and demeanor, 2) which ones rank highest given the remaining attributes.

Since demeanor and genetic diversity are difficult to fake long-term, the maximizing Pairer should fake as many of the other signals as possible and find as many other candidates as possible.

Well, that's the theory and practice, at least. ;)