Theory & Writings

Alimony (The Program): Some Thoughts/Explanations

Craig Ross

Ann Arbor, Michigan

December 1995 (Revised August 1998)

Why anyone would want to hear the "logic" behind Alimony is beyond me. My questions are always result-oriented. Does it make sense? Does it piss me off? I have no inclination to grapple with the intricacies of my PC or television (or corkscrew for that matter) to discern the "how" of things. The question I have is: will it open a decent bottle of Cabernet without pieces of cork settling in? Well, I know, we all come from different consciousness. My personal idol is Juwan Howard, a former basketball player at the University of Michigan (and now coach at the University of Michigan), who upon being questioned (many years ago when he was a college player) about the merits of the beaches in Hawaii, said he wasn't impressed, because "there was too much sand and shit." My sentiments exactly.

1. Ordering of Claims.  

The "logic" of Alimony springs from a seminal thought that inspired my 1979 first draft -- that it makes sense to order claims on a continuum of strength. Maybe there is a more creative/holistic vision, but I can't figure out how to start at the problem until cases are "organized" by some vision of "merit." Regardless of how one gets there, I would suggest that some kind of a "sort" theory is necessary in any attempt at applying numbers (algorithms) to the history of a marriage. 

The worst possible claim in Alimony is designated 0, the best 100. These poles are (a little) more than this, since I would label them circumstances where (the 0 spot) "everyone agrees" the assertion is pathetic, and those where (the 100 designation) "everyone agrees" permanent alimony should be a lock.

The shortest distance between two points is a straight line. Between those a place where someone is always throwing you a curve. There is no mandated reason, of course, for a continuum to be linear as opposed to "curved." Indeed, the possibilities for "drawings" between point 0 and point 100 are infinite. However, I have elected a "capped-at-the-end" line model since (a) linearity is easier to comprehend, (b) trial and error testing of the theory is simpler, and (c) a straight line, as an ordering mechanism, tends to place close cases "close" and distinct cases "far apart." In short, a straight line accomplishes the stated goal.

Consider, for example, on-base percentage in baseball. We could define everyone with an on-base percentage of .280 and lower as (say) "Felix Fermin" or "Chico Lind" and everyone over .390 as "Frank Thomas" or "Jose Canseco." But it is useful to know that Chico, at .275, has greater utility than Felix's (the dog) incredible .239. Similarly, Canseco, while good, does not hold a candle to the Big Hurt. On the other hand, the difference between Lofton and Belle (Dybzinski and Mendoza) may not be worth talking about. The Dibber and Mendoza had no point in the Bigs. Lofton and Belle are (future) Hall of Famers with relatively equivalent values. Hence, the "capped" line. A marriage of 40 years, for alimony purposes, is no different from one of 42 years. Ruth versus Gehrig. Dick Stewart versus Marv Throneberry. Bud Selig versus Elmer Fudd versus Harry von Zell. Rush Limbaugh versus Strom Thurmond.

The components of any such a line represent value judgments. The one I define has two (primary) elements.

The first is "need" in relation to the ability of the claimant to support herself/himself. We live in a dog-eats-dog-and-every-man-for-himself-and-god-against-all society where individuals are compensated in relation to the scarcity of their skills in the marketplace. Alimony attempts to measure the claimant's present abilities and potential abilities in this kind of environment.

The second element is the disutility of the marriage vis a vis labor market ability/potential. It may be that a claimant has no present ability to earn, but the marriage (short duration, no kids) has not had any impact (or marginal impact) on that condition. Or it may be that the claimant once had a labor market future, but the duration of the relationship and his/her role in the marriage has made this potential problematic or nugatory.

The point of the continuum in Alimony is to weigh both questions – need and disutility. Which reminds me. Bill James once asked himself the following question: "Can Jose Mesa pitch?". His answer was: "Can a bear perform brain surgery? I guess we'll never know till we hand him a scalpel and stand back." We now all know that bears make great brain surgeons, particularly if one can define the correct (one inning) operations to perform.

Alimony asks six questions in measuring need/disutility. These questions/criteria are: (a) length of the marriage (b) income (actual or construed) of claimant (c) education of the claimant (d) age of the claimant (e) number of children born to the couple (f) other party's income.

While, at first glance, this may appear a fairly limited list, I would argue (absent some other suggestion yet to be made) that the foregoing criteria pretty much cover the field. No court, genius, or bear anointed brain surgeon can extract the day to day nature of a marriage. However, we can predict, using the claimant's present income, age, and education, what he/she can expect in a short- to mid-run foray into the marketplace. And we can infer, from the duration of the relationship and the children born to the couple, the probable cost to the care taking spouse.

This leads to three questions.

First, one might ask if income differential is a part of the mix. I struggled with this one in 1979, but the consensus of the (many) lawyers I consulted was that it should be; that "need" is relative to the historical standard of living in the family. I have a low view of consensus, but in this instance I endorse the "majority" view as logical and imperative. Consider, for example, a couple married for 30 years, each in their 50s. They have raised 3 children (now grown) and have little in the way of savings or assets. Assume the claimant earns 40k, the other party 300k. Would we say, in this circumstance, that the claimant's 40k is "enough" -- that a single person in our society can survive at this level? Or should it be that the "partnership" has generated an asset (future cash flow) subject to a degree of sharing, and that mere survival is not the end of analysis? there is no "objectively right" answer to this. It just depends upon your politics and world view. I lean away from the pure capitalist (dog-eat-dog) view. Alimony adjusts income differential by scoring the income component of the line "higher" as the parties' incomes diverge, at least beyond certain levels.

A second question might relate to the inter-correlation of the "age" and "length of marriage" factors. There is, of course, inter-correlation, but the variables are weighted to minimize any synergistic effect. Moreover, these factors do not always "run together." Some people marry later in life, and (in addition, aside from this) it is logical to consider a claimant's age, since success in the labor market is related to this reality, particularly where an individual has been out of the work force, or has marginal skills.

A third question might be, "what about the law?". Just like a lawyer to be concerned over the ravings of some bowling team that hasn't thought about alimony for more than 5 seconds, except if he/she has got to shell some out and is pissed. In point of fact, the various courts have defined criteria that tend to mirror (a) through (f) above, or, at least, (a) through (f) encompass appellate concerns. The cases talk about age, duration of the marriage, "station in life," health, the ability of a party to support himself/herself, (blah, blah, blah), considerations all manageable within the Alimony criteria.

The cases also talk about "fault," a notion I have excluded. Two basic reasons for this. First, I don't know how to quantify it. I don't know how anybody does. I don't believe anybody can. Courts ain't god and me neither. ("My dog likes apples and oranges too, That's how we're alike"...Rod McKuen. "My fish likes fish food and I don't, That's how we're different"...Roger Ross.) If fault was included into a theory, there would be no value to any ordering of cases or, for that matter, any attempt at quantification. All cases, and the same case, would be a 0 or 100 (pathetic or great) depending upon whether the evaluator was claimant's or respondent's attorney. Border wars would transform to borderline personality defect wars. 

Second, I believe the focus on "fault" is erroneous, a false issue that has little or nothing to do with alimony or anything else that goes on in domestic relations cases. Excluding of course, instances of frank abuse, where guilt or criminal liability are at issue, rather than "fault." And this conclusion assumes "fault" is possible, a condition I deny. This is not to say that betrayals or misbehaviors do not occur. In my experience, however, these transgressions are not easy (possible) to sort or ascribe with any economic meaning.

      He had an affair. Yeah, but she wouldn't sleep with him. Yeah, but he smelled bad. Yeah, but he wouldn't smell so bad if she wasn't so cold and such a nag. Yeah, but she wouldn't have been such a nag if he didn't cruise with the Tooz at the Doo-Dah Room. Yeah, but he wouldn't cruise if she would cook a decent meal for him.

I just do not believe any court or person can discern fault, sort through these normal/lunatic pronouncements, even if such behaviors have any legal or economic meaning. Most of the time the marriage just sucked or was a bad match of egos. And the other times no third party can tell, and the litigants probably don't really know. "Fault" litigations are the lowest possible path, probably how Mark Fuhrman or Hannibal Lecter would design a legal system, a way for lawyers to earn money under false pretenses. To a great extent, the concept is at fault for the battering the profession receives. And don't tell me about the standard of practice. Stop me before I kill again, and fault is for someone else's memoirs.

As a matter of numbers, the segments of the line are --

1. Duration

0 - 35

(0 - 30 years)

2. Income

0 - 30

(0 - 40,000 dollars)

3. Education

0 - 15

(none to M.D. degree)

4. Age

0 - 10

(30 - 60 years)

5. Children

0 - 10

(0 - 3 kids)

Note that "Income" is a number that is adjusted by income differential, and "Children" has a possible negative (-5) component. Again, there is no magic in these comparative weightings. After a long trial and error process, the foregoing were found to result in an ordering of cases that made sense to me. Usually not much cork in the wine. But I view the relative impacts of the criteria to be an open question. It is probable I will make minor adjustments over time; and it is a certainty that I will "index" the "income" numbers and continue to tinker with the income differential algorithm. This is all the result of a lot of trial and error, and looking at answers that made sense to me.

2. Impact of the Order.

OK, now every case can be assigned a number between 0 and 100. By definition 0 is "absolutely no case" and 100 is "a lock for permanent alimony." The scores in between, as they move from the poles, become more subject to individual impression. Again, applying a result to a score was a matter of many samples, thousands of hypothetical runs. First, two general scales were created. These are—






No Case


No Case


On the Margin


On the Margin


Strongest Case


Strongest Case

The scale on the right is applicable only when the parties have an income differential of $30,000 or more. This may create a minor "cliffing" problem. However, since the cases in the range are only designated as "marginal," the anomalies are rare -- for example, a $29,000 income differential in an (otherwise) weak case. I may continue to make some minor adjustment here in future versions.

The scales are subdefined and the "breaks" in categories are arbitrary/trial and error but not critical in that the program recognizes marginal differences as just that, and takes no special account of, say, a case scoring at 73 versus 74. The subdefinitions are as follows --

Under Margin (25 or 42)

No alimony

At Margin (25-50 or 42-50)

Possible Award

Short Term Award (50-66)

60 Months or Less

Long Term Award (60-83)

72 Months or More

Permanent Award (73-100)


This was organized by instinct/trial and error. There is no science in the foregoing categorizations. Nor can there be. The results are what I think makes sense, with the caveat that any equation will sometimes produce a result that I (for one) might question. The same would be true for any other attempt to create numbers out of non-numbers. Note the overlap in the brackets, diminishing "cliffing" problems and anomalous results.

3. Amount of Award: No Minor Children.

The assumption of the program is, in the strongest possible case, that the parties will share incomes. In other words, in a long term marriage (with no minor children) designated as "100" the parties will retain equal cash flows. As the case moves from the strongest to the "marginal" area, the sharing becomes less pronounced -- more homage to atomism (capitalism) and less to Engels. Again, the deviations are made along a straight line. The formula (F1) is --

Alimony = (Score divided by 100) times half the difference in the parties' incomes, where gross income has been adjusted for FICA/SE.

In that FICA/SE cannot be shifted, the adjustment is made prior to an "answer" calculation. Thus, if a claimant's FICA/SE adjusted income is $20,000 per year, and a payor's FICA/SE adjusted income is $50,000 per year, and the case scores out at 70, alimony would be --

      (70/100) * .5 * ($50,000-$20,000) = $10,500 per year.

The primary deviation from the foregoing is what I have labeled the "mailbox" postulate. This is, in part, the inverse of the notion of "income differential" (as justifying an award). I would argue that the difference in litigants' incomes is relevant to an analysis of the strength of a claim, but not (at least necessarily) to setting an award as based upon the entire level of the difference. This is because (a) there are (very significant) costs to working as opposed to not working (e.g., taking care of one's house, travel time to work); (b) there may be costs to working a high-profile/high income versus low-profile job (e.g., often many more hours worked, wardrobe); and (c) it is harder to get up every day and hunker down to work than it is to straggle out to the mailbox. Admittedly, (c) is a (crass) oversimplification involving a limited number of cases, but it is a part of most all of our consciousness, an American consciousness (if a passing one). And this is particularly so in cases involving extremely 18pt incomes/income differentials.

We all have some sense, even if it is a 12pt sense, that it is a winner-take-all game. Or, at least, that a person deserves some "edge" for the effort of success. Thus, we all hate the IRS and flirt with the moronic concept of flat taxes, and do so even if it is irrational and even if the money were to be spent exactly as we designated. The USSR did not break up over ideological diversity. The truth is the party went out of bounds because nobody wanted to wear Bulgarian shoes or jeans stitched in Latvia. Give me those USA Levis and Italian Continentals (but not size 12, Bruno) and Dick Vitale, baby.

The adjustment is made by adding an additional multiplier to Formula F1 above. Up to $20,000 in difference is not adjusted (or, the multiplier is 1). At $290,000 or more per year in difference, the multiplier is the minimum, .698. Between $20,000 and $40,000, the multiplier decreases from .98 to .92, in four equal intervals. Between $40,000 and $90,000 it is .91. Between $90,000 and $290,000 of income differential it is --

      Multiplier = .91 - .212 * (Adjusted Income Differential less $90,000 divided by 200,000)

"Adjusted income differential" is a reference to post-FICA/SE dollars. This is another trial-and-error arena and one where I continue to run numbers. Assume a payor earns $350,000 (post-FICA/ SE) per year and a payee is imputed with a $20,000 adjusted gross. It is a long marriage and the case scores at 80. Without the "mailbox" adjustment the alimony amount is --.

      (80/100) * .5 * ($350,000 - $10,000) = $136,000 per year

With it, it is --

      (80/100) * .5 * .698 * ($350,000-$10,000) = $94,928 per year

Now, of course, if one actually saw the people and knew the course of their relationship, the properties to be owned, responsibility toward adult children (and on and on), the first answer (or more) might seem right. There are an infinity of possibilities. On the other hand, as a completely abstract proposition, most people I have consulted feel more comfortable with the latter number. So do I, though I am insulted by neither. I guess the question is, would you rather have 95k and not go to work or 255k and go to work? My suspicion is there are people who would choose neither, would prefer the 95k and work, and then the answers would be all over the board. But I don't really know. I perceive this as an area of continued tinkering.

4. Amount of Award: Minor Children.

Similar to section 3, above, but, as opposed to a 50/50 sharing at the 100 pole, the skew is adjusted in accordance with the parties' direct responsibilities for children. Assume a three-person family becoming 2-person and 1-person households. If it is theorized that economies of scale do not exist, that 2 people cost twice as much as 1, the two-person family, to be at the same spot as the one-person household, needs 66.66% of family resources. If economies of scale are pristine, that is, if it is the case that 2 people cost the same as one, then a financial equilibrium results via a 50/50 cash flow split. Neither case is fact, but the idea defines the poles of possibility in a three-person family and a classic custodial/non-custodial restructure. Which pole is likely to be closer to reality? My guess is "neither one" or, "about some place in the middle." 

In other words, economies of scale are a reality and that reality is about "half," in relation to the paradigm. This would imply, in the example, that the 2-person household needs 58.33% of family resources for equilibrium to exist. In fact, the Department of Agriculture has conducted studies (Consumer Expenditure Surveys) that premised the creation of charts (Comparative Equivalence Scales) comparing various family compositions in relation to intact 2-adult/2-children households. From these studies the costs (equilibrium) between households of varying residents can be calculated.

In the 1970s a University of California economist, Philip Eden, wrote a book, Estimating Child and Spousal Support, demonstrating the arithmetical manipulation of the scales. Dr. Eden also argues that the comparisons should be adjusted for housing cost, not a factor in the source data (the federal surveys). Using the charts (as redefined in the treatise) he calculates an equilibrium (in the present hypothetical) of slightly less than 59% -- just about what one's (economies of scale) instincts would suggest. Alimony adjusts the percentage based upon the family composition and (if shared or split) the custodial arrangement.

In relation to a three-person, ordinary custodial arrangement the alimony formula becomes the same as F1 above, but with the "pristine" case (100 score) retaining 55% of family cash flow. It looks like --.

    Alimony = (Score/100) * (Mailbox Adjustor) * (Total Family Income After FICA/SE Adjustment * Economies of Scale Percentage) - (the Payee's Post-FICA/SE and Child Support Income);

     where "EOS" percentage (in this case) is .55

Two questions arise out of this.

The first is, why use 55% as opposed to 59% in adjusting the equilibrium between households? I admit that 55% is a guess, but my surmise is that Eden understated housing (quasi-fixed) cost in his calculations and ignored a degree of non-custodial cost sharing. I could be wrong, but the variance (in any event) in the range is pretty 12pt in most cases. It is a lock the answer is between 55% and 60%, and I chose a conservative line.

The second question is, how does the equation work as to child support, when child support (unlike alimony) is payable in net dollars? The answer is that Alimony adjusts child support back into its gross equivalent. In the past I had suggested the process move the other way (gross to net), but this proved unwieldy where the answer sought was in alimony (gross) dollars. The program uses an approximate conversion, based upon the parties' income levels. In that I chose conservative "family composition" percentages (and since it is the payee's household that is focused upon), the payee's tax rate is used in making the transformation to gross dollars. (More detail available on request.)

The theory/process outlined above is repeated in Alimony for other family compositions and arrangements. This includes a sliding percentage in joint and split custody cases depending upon (split custody) "where are the kids?" or (joint custody) "how much time does the child spend in each household?"

5. Unallocated Support.

Alimony will suggest an alternative to child support and alimony in the form of a "unified" family support calculation. The formulas are rudimentary and not "adjusted" so as to appear (necessarily) "harmonious" with the alimony recommendations made by Alimony, but, to this point, they work pretty well. Continuing review of this aspect is probable in future versions, but I see so few of these cases (since the reduction of federal tax rates at higher income ranges) that I am not going to get real worked up about it.

6. Duration of Awards.

Two factors are most salient in determining the duration of any alimony transfer. The first, axiomatic, is the strength of the case. The second, a subset of the first, is the number of years of the marriage.

The starting spot for the duration numbers is the chart shown in the Alimony 2.02 pamphlet. If the case scores over 83 (level one), a recommendation of permanent alimony is made. If it scores at level six, it is suggested alimony be barred. Thereafter, the ranges are measured by assigning a duration as a simple (sliding) multiplier against the years of the marriage. A partial list is outlined --


If the marriage is more than 20 years, Alimony suggests that a permanent award is possible. Otherwise, the duration is .45 times the marriage term (L). If the marriage is less than 20 years, the possibility of a permanent award is not suggested.


Duration is between 72 months and .395 * L years


If the length of the marriage is greater than 17.8 years, the range is 72 months to .35 * L. (Note that .35 * 17.8 = slightly more than 72 months.) If the length is less than 17.8 years, the suggestion is between 60 months and .3 * L years.

Why these numbers? Aren't these just arbitrary judgments? The answers are "sand upon sand" and "yes." But you to have to keep your eye on the original question. "How much alimony" and "how long should an award run" are purely value judgments in the first place. There is no way to make a science of the answers. All one can do, in my opinion, is create a theory, run thousands of trials, and then adjust the formulae (and run more trials) until the results "look right." What "looks right," of course, is tempered by individual bias, whether the market place of marital relationships is "caveat emptor" or can/should be something else.

The criticism I am most interested in relates to specific fact situations. If a practitioner uses Alimony and finds a result to be out of bounds or anomalous, I am interested in seeing the case and any explanation for why the result is "wrong." Oddly, despite the fact I have distributed hundreds of guidelines to law firms throughout Michigan, I have yet to receive any commentary/criticism to this end.

I continue the process of self-criticism in my own caseload, but I never hear much beyond these cases, save for the occasional rumor that it is "too high" or (less usually) "too low." Aside from this, I remain interested in comments, particularly "technical" comments about "how" a guideline should be structured.

Back to Top