SILLS, P.J.—In this case we hold that the trial court abused its discretion by calculating income for pendente lite child and spousal support orders based on only the latest two months of a salesperson’s earnings. The trial court should have used a properly representative sample, which, on this record, would presumptively have been the previous 12 months.
1. The Sample the Trial Court Used
Husband Tracy Riddle works as a commissioned financial advisor for a major investment firm. There are four components to his income, the first two of which are somewhat unusual in that they constitute (1) the forgiveness of • debt on an advance received by his employer during the marriage and (2) the forgiveness of the interest on the debt.
The circumstances behind these two components are these: During the marriage, in order to lure the Husband away from his prior employer, his present employer paid him an “advance” against his future earnings of $1.039 million. This amount was to be paid back every month in installments of $11,423.14, plus interest of $2,365.54. However, on Husband’s earning statement the $11,423.14 plus $2,365.54 is included among Husband’s earnings because the amount is also being forgiven in the same monthly installments in which it was supposed to have been paid back. (In tax law, the forgiveness of debt is considered income.) The reason for this convoluted system of payment is fairly obvious: It allowed the employer to pay Husband big bucks up front, but spread out the payment of tax on the payment over time so as to circumvent the progressivity of the tax codes.
There are two other components to Husband’s earnings as shown on the employer’s earnings statement besides the $11,423.14 forgiveness-of-debt and the $2,365.54 forgiveness-of-interest on the debt. One is (3) a monthly draw of $2,340. The other, and this the variable part, is (4) the actual amount of commission Husband earns for a given month, and that figure fluctuates.
Wife Susan Riddle is, by her own description, a stay-at-home mom with no income. At the initial order to show cause for pendente lite child support and spousal support on February 28, 2003, the trial court made an order that Husband pay $3,619 per month in child support, plus 16 percent of any income in excess of $21,950 in a month. It also ordered that he pay spousal support of $4,338 plus 20 percent of any income in excess of $21,950 per month. Both orders were predicated on a finding that Husband’s gross monthly earnings were $21,950. Husband timely appealed from that order.*N1
The $21,950 was arrived at by taking the combined figures on Husband’s year-to-date earnings for 2003—that is, just the two months of January and February 2003—for the draw, the commission, the forgiveness of the $11,423.14 monthly installment to repay the principal on the advance, and the forgiveness of the $2,365.54 installment on the interest. For those two months the total of these amounts was $49,789. The trial judge then divided the • $49,789 by six weeks, apparently because the last entry on the statement was as of February 15. (Six weeks from January 1 through February 15.) That number was $8,298. Then that figure was multiplied by four to obtain a “monthly” income of $33,192.
However, the court then deducted $11,242—obviously an erroneous transposition from the $11,423 principal forgiveness figure—and arrived at $21,950, which was then plugged into an “Xspouse” software program (apparently Xspouse is a new rival to the dissomaster). The Xspouse software then yielded two figures for child and spousal support.
The trial court’s deduction of the $11,242 has not been challenged in this appeal, and there is authority (albeit in the form of dicta) for it, in In re Marriage of Kirk (1990) 217 Cal.App.3d 597, 607 [speculating that if repayment of debt arrangement was not truly “‘voluntary,'” trial court would have discretion to deduct it from income on remand].*N2
2. Alternative Samples the Trial Court Might Have Used
The $21,950 contrasts starkly with the figures the court would have obtained if it had used a more representative time period than just the first two months of 2003. The combined figure for the draw, commission, and forgiveness of $11,423.14 for the previous 14 months was $266,197.63, which would amount to an average of $19,014.12 a month. Assuming the $11,423 was deducted from this average just like the trial court had deducted it from the $33,192 figure, the result would be $7,591.12 — which is only 34 percent of the $21,950 attributed as income by the trial court.
Or, if one took the 12 calendar months of 2002 as a more representative figure, the combined totals would have been $216,408.63, or $18,034.05 a month. After deducting the $11,423 forgiveness of principal the monthly income figure would be $6,611.05. (Which is about 30 percent of $21,950.)
Finally, if one took just the last 12 months prior to the February 28 OSC—a combined total of $237,808.59 (which included amounts deducted for January and February 2002, i.e., $28,388.44, but with the $49,789 for January and February 2003 added back)—and divided that by 12, one would get $19,817.38, and after the $11,423 was taken out, the total would be $8,394, or about 38 percent of $21,950.
A. Cash Flow Is Not Necessarily “Income”
At the outset we must reject Husband’s argument that the only substantial evidence at the hearing necessitated a finding that his income was a mere $1,700 a month because that was what his expert testified his “cash flow” was, and there was nothing to contradict it. While we recognize that family lawyers and forensic accountants sometimes use the phrase “cash flow” as a sloppy synonym for the word “income” as it appears in the support statutes, it isn’t. In particular, the child support laws (see Fam. Code, §§4058, 4059) are very exacting as to the definition of income.
As we pointed out in In re Marriage of Schulze (1997) 60 Cal.App.4th 519, 529, the language was “lifted” straight from the Internal Revenue Code. That means that if the tax laws say you have income because of the forgiveness-of-debt, you have income, and that forgiveness-of-debt income must go into the calculation of adjusted gross income under section 4058, subdivision (a), which in turn is the basis for income under section 4059, subdivision (a). Now, there is authority in In re Marriage of Kirk, supra, 217 Cal.App.3d at page 607, as we have alluded to above, to ameliorate the harsh effects of assessing “phantom” income as imputed by the tax laws under the circumstances of a given case (e.g., §4059, subd. (f.)), but that doesn’t mean that so-called “phantom” income as imputed by the tax laws is any less “income” for purposes of section 4058, subdivision (a).
In this case there was evidence to contradict the expert’s testimony as to Husband’s income being a mere $1,700 a month. First, there are the figures from the employer’s earnings statement that we have already discussed. Second, in addition, there was also the Husband’s W-2 form for the calendar year 2002, which showed “wages, tips and other compensation” of $182,920.12 and “Medicare wages and tips” of $202,229.15. (Note, however, that even using the higher of these figures, the monthly average would be $16,853, and assuming the $11,423 deduction for the earlier loan, the figure would be $5,429, which was lower than any of the alternative sample calculations set forth above.) Though the W-2 contradicted Husband’s income and expense declaration (which asserted net monthly disposable income of $2,197), the court would have been entitled to consider the W-2, as well as the employer’s statement, better evidence of what his real income was in February 2003. (See In re Marriage of Loh (2001) 93 Cal.App.4th 325, 332 [presumptive correctness of tax returns].)
So the theory that a $1,700 a month cash flow should necessarily be equated in this case as the Husband’s “income” simply is not tenable.
B. Time Samples Must Be Fair and Representative to Determine Fluctuating Income
But just because the Husband has taken an overstated position by equating cash flow with income doesn’t mean the rest of his appeal is without merit. Husband is quite correct that the trial court used an unjustifiable sample.
As regards sampling, this case is essentially a gloss on Family Code sections 4060 and 4064, the two statutes which deal with the problem of calculating fluctuating income for support orders.*N3 The aim of section 4060 is to give a trial court discretion to adjust the annual net adjustable income required for a support order when dividing net disposable income by 12 “does not accurately reflect the actual or prospective earnings of the parties.” Section 4064 gives a trial court the authority to adjust a child support order “as appropriate to accommodate seasonal or fluctuating income of either parent.” While both statutes are framed in discretionary terms, it is also well established that the discretion must be a reasonable one, “‘exercised along legal lines, taking into consideration the circumstances of the parties, their necessities and the financial ability of the [supporting spouse].'” (In re Marriage of Rosen (2002) 105 Cal.App.4th 808, 825, quoting In re Marriage of Laube (1988) 204 Cal.App.3d 1222, 1225, bracket insert made by the Rosen court.)
The published case law articulating the basic common law principles as regards fluctuating income is still reasonably manageable at the moment.
From In re Marriage of Hall (2000) 81 Cal.App.4th 313, we learn that a court must arrive at a stable number in order to make a support order, even if income does fluctuate from month to month. (See id. at pp. 317-318 [“The formula is always predicated on knowing what both parents’ income is in nominal static dollars at the time the order is made.”].)
From County of Placer v. Andrade (1997) 55 Cal.App.4th 1393, we learn that the idea behind this stable number is to have a reasonable predictor of what each spouse or parent will earn in the immediate future. (See id. at p. 1396 [“The assumption underlying these calculations is that past income is a good measure of the future income from which the • parent must pay support.”].) The theory is that the court is trying to predict likely income for the immediate future, as distinct from extraordinarily high or low income in the past. And, perhaps most relevantly to the case at hand, In re Marriage of Rosen, supra, 105 Cal.App.4th at page 821, has recently demonstrated, albeit in the context of calculating the good will in a small law practice, that the time period on which income is calculated must be long enough to be representative, as distinct from extraordinary.
In Rosen, a panel of this court considered the case of a lawyer whose income varied dramatically in the four years prior to his 1996 divorce. In 1992 it was down at about $73,000, in 1993 it was up to about $101,000, then in 1994 it was down to about $71,000, but in 1995 it was way up to about $140,000. The lawyer’s ex-spouse presented expert testimony basing the goodwill value only on the (historically abnormally good) year of 1995, and the trial court bought the expert’s valuation based on that higher figure.
The selection of the unrepresentative year was error, said the Rosen court. The expert should have “averaged” the figures rather than just taking the latest figure in a transparent effort to inflate goodwill. (See Rosen, supra, 105 Cal.App.4th at p. 820 [“Under the facts presented here, we believe Pat’s expert ‘should have averaged it.’ . . . A reasonable trier of fact could not help but conclude the expert chose to use Bruce’s net income from 1995—one of Bruce’s highest earning years—solely to inflate the value of goodwill.”].) The income for 1995 was not an average or “‘reasonably illustrative'” of the lawyer’s earnings. (Ibid.)
Rosen involved a goodwill calculation, hence the focus of the court was on average annual income. The present case, by contrast, involves a support order so the focus, as we might deduce from County of Placer, is not on long-term prior average earnings calculated over many years in order to obtain a reliable figure, but on immediate prospective earnings on which support must be paid. Even so, the basic principle which Rosen applied to goodwill is also soundly applicable to support orders. Indeed, it is a principle familiar to anyone who has ever read Darrell Huff’s 1954 classic, How to Lie With Statistics (often used as a textbook in high school critical thinking courses), or had a course in elementary statistics or logic: A sample must be representative of what is being sampled.*N4
So what was implicit in Rosen as regards a goodwill calculation we will now make explicit as regards to support orders: It is a manifest abuse of discretion to take so small a sliver of time to figure income that the determination essentially becomes arbitrary. And under the facts of this case, no other word but “arbitrary” properly describes the trial court’s selection of the last two months to determine Husband’s income. A mere two months is an embarrassingly short period on which to predict the annual income of a commissioned salesperson who works in the financial markets. And it is particularly too short under the record before us, where Husband’s income for the previous 14 months, previous calendar year, and immediately preceding 12 months were all (a) pretty consistent in themselves ($7,591.12, $6,611.05, and $8,394, and if you used W-2 medicare wages, $5,429) but (b) wildly inconsistent with the two month period January–February 2003 ($21,950).
We are forced to conclude that the trial court chose an unrealistic time sample in order to, as the Rosen court might have put it, “inflate” the supporting spouse’s monthly income.
Since both the child and spousal support orders were predicted on the (unrealistic) $21,950 monthly income figure, they must both be reversed. The question then arises as to how much guidance we should give the trial court on remand. Since section 4060 and 4064 are framed in discretionary terms, it would be outside the proper province of an appellate court to prescribe a bright-line rule for the precise parameters of a proper sample; after all, the whole point of discretion is a recognition that there are times when there shouldn’t be a bright-line rule.
We also recognize, of course, that, as in Rosen, longer time samples are appropriate for measuring “average annual income” for purposes of goodwill valuations. However, as regard support, we may say that statutes appear to create a presumption that the most recent 12 months is certainly an appropriate period in most cases.
While section 4055 is framed in terms of “net monthly disposable income” (see §4055, subd. a)), section 4060 defines that phrase in terms of “dividing the annual net disposable income by 12” and section 4059 defines annual net disposable income in terms of “annual gross income.” In short, there is a heavy emphasis on 12 months or “annual” income as a benchmark for the • calculation. Which is, come to think of it, only
common sense anyway: The income tax laws which, like support orders, are focused on income, are also framed, not in terms of artificially truncated and therefore unrepresentative slices of time, but in terms of whole years.
A longer period could, of course, be conceivably used under section 4060 if it were more representative of a party’s income. For example, a two- or three-year average might indeed yield a more representative overall income figure for a party who only wrote books for a living. One can readily imagine that in the book industry long production lead times and the tendency of initial sales to fall off (hence royalties are likely to be highest with a book’s initial release) would mean that a period of more than a year might be necessary in order to obtain a representative picture of an author’s income.
Then again, a longer period could well have its problems in regard to a salesperson (as distinct from an author), particularly a salesperson who worked in an industry that traditionally bobs and dips with interest rates, or the economy as a whole (e.g., real estate sales, or, as in the present case, securities sales). Going back too far for such income earners only yields figures reflective of what the overall economy was in the past, not a figure, as the statutes and case law contemplate, of what the party’s income will be in the immediate future. In the present case, for example, going back to the late 1990’s might only show what a commissioned investment advisor could make during the “dotcom boom.” That merely shows what was “likely” back in 1999, not what was likely in February 2003.
On the other hand, too short a period (and the case before us practically represents a reductio ad absurdum for too short a period) clearly exaggerates events in that period. Here, Husband obtained a whopping large commission in February 2003, and the trial court predicated its order on the unrealistic theory that such commissions, clearly out of line with what he was making in just the previous 12 to 14 months, would continue indefinitely. We have only to flip the facts around to see the speciousness of the analysis: Had the Husband had a lean two months in the period January through February 2003, an order predicated on just those two months would have unrealistically understated his income. Additionally, time periods that are too short may unrealistically exaggerate economic micro-trends that may, over longer periods, smooth themselves out. Surely anyone who has even casually followed the Dow over the past decade knows that there can be periods of many months of up followed by many months of down—over the course of a year a trend may emerge, but slicing just one small period of a few months is likely to yield a highly unrepresentative picture.
C. This Order Cannot Be Justified on the Basis of Imputed Income
The court’s math and statistical errors*N5 were brought to the trial court’s attention by formal objections to the court’s tentative ruling.*N6 In a hearing on the objections the court alluded to the idea that income should be imputed to Husband based on his historical earnings.*N7
The simple answer to this issue is that there is nothing in this record on which to predicate an imputed income award.*N8 This is not a case like In re Marriage of Destein (2001) 91 Cal.App.4th 1385, where a spouse who never had a real job since the marriage (given that he had sold his businesses before the marriage) and had sizable separate but illiquid holdings in real estate from which a return could be imputed in order to ascertain income. (See also County of Kern v. Castle (1999) 75 • Cal.App.4th 1442 [large inheritance considered in setting support level].)*N9
Nor is it one of the variations on the theme of deliberate underemployment on which cases have predicated imputed income. (E.g., In re Marriage of Hinman (1997) 55 Cal.App.4th 988 [income imputed to noncustodial parent with degrees in computer science on theory that she was eminently employable in computer field]; In re Marriage of Padilla (1995) 38 Cal.App.4th 1212 [parent resigned his current employment to start business which, at time of hearing, was not generating any income at all]; see also In re Marriage of Regnery (1989) 214 Cal.App.3d 1367, 1376 [under bad faith rule for imputing income, finding deliberate refusal to obtain employment where parent quit job as senior cost accountant because he found it too burdensome and limited subsequent job seeking efforts to cost accounting/auditing field].)
In that regard, we must reject the idea that just because there were years in the late-1990’s in which Husband earned about $300,000 as a commissioned investment salesperson (which could, if you do the math, yields a $25,000 monthly income) Husband could earn the same amount in early 2003 when this order was made. This is the logical fallacy of extrapolation, in which some series of events in the past is necessarily assumed to continue in exactly the same way into the future. Pretty much every sentient adult in the United States knows that the stock market did not continue to go up in the early 2000’s the way it had in the late 1990’s.*N10 The idea that a salesperson will necessarily do in the immediate future what he did in his or her best of recent years is both illogical and contrary to Rosen’s principle that samples must be representative, not selected to skew results one way—or the other.
The pendente lite orders for child and spousal support made by the court are reversed, with directions to make new orders based on net disposable monthly income as shown by this record,*N11 calculated not inconsistently with the views expressed in this opinion.*N12
In the interests of justice Husband will bear his own costs in this appeal.
Bedsworth, J., and Moore, J., concurred.
n1 Wife has not filed a brief. However, we do not treat the failure to file a respondent’s brief as a “default” (i.e., an admission of error) but examine the record, appellant’s brief, and any oral argument by appellant to see if it supports any claims of error made by the appellant. (See Cal. Rule of Court, rule 17(a)(2); In re Bryce C. (1995) 12 Cal.4th 226, 232-233.)
n2 It is dicta because, strictly speaking, Kirk only held that under the facts before the court, it was an abuse of discretion to deduct the repayment of the loan; the court allowed, however, that, under different circumstances, it might not have been an abuse of discretion to deduct the loan repayment.
n3 Family Code section 4060 provides: “The monthly net disposable income shall be computed by dividing the annual net disposable income by 12. If the monthly net disposable income figure does not accurately reflect the actual or prospective earnings of the parties at the time the determination of support is made, the court may adjust the amount appropriately.”
Family Code section 4064 provides: “The court may adjust the child support order as appropriate to accommodate seasonal or fluctuating income of either parent.”
n4 One of history’s best examples of how using a non-representative sample can yield disastrously wrong results is the famous Literary Digest poll for the 1936 election. The publication predicted a 15-point landslide for Alfred M. Landon over Franklin Delano Roosevelt. In reality it would be Roosevelt who would win in a landslide. (See, e.g., Jacoby, A Critique of Rappeport’s ‘Litigation Surveys—Social “Science” as Evidence’ (2002) 92 Trademark Rep. 1480, 1495.) The Literary Digest used for its poll a sample of voters drawn from telephone and automobile registration lists, i.e., people who had phone service and owned automobiles, a group which, during the Depression, was disproportionately wealthy in relation to the general population. (See ibid.) The poll was disastrous for the Literary Digest, which soon went out of business, becoming a virtual by-word for poor sampling.
n5 For example, as alluded to above, the trial court divided the figure for the first two months of 2003 by six instead of eight because the earnings statement’s last entry was February 15, even though there was no evidence that Husband would receive any more earnings in February. That further exaggerated Husband’s income by yielding a number significantly larger ($8,298) than division by eight ($6,223).
n6 The ruling was announced by way of a tentative ruling filed March 28, 2003. The January 16, 2004 notice of appeal is from the formal orders on order to show cause filed on November 17, 2003, embodying the same orders as those tentatively announced on March 28, so the appeal, involving a matter within the traditional collateral matter exception to the one final judgment rule, is timely.
n7 Here are pertinent parts from the transcript of the hearing on the objections:
“The Court: I understand. I understand, Mr. Schilling, but this is a gentleman who’s making $300,000 a year historically. And rather than just picking a number to impute from the history, I determined that I would use some number basis taking into consideration the money he has to pay back to the individual [apparently referring to the payback of the advance]. [¶] . . . [¶]
“Mr. Schilling: Well, with the facts being that while historically the man did make, you know, $300,000 a year, but when he changed jobs, they gave him in effect an advance on that, which is now gone because the parties spent it, now he’s having to pay that back, it puts him in a position where he has no capability of complying with the order because—
“The Court: Well, counsel, I did not hear any evidence that he is impaired mentally or physically from performing his job. He has a history of substantial earnings.”
n8 An issue which has not yet been resolved is the degree to which a full-time childcaretaker should have income imputed to him or her on the theory that, after all, he or she could put the kids in daycare and go out and get a job. (See In re Marriage of Cheriton (2001) 92 Cal.App.4th 269, 301-302 [suggesting that such imputation would be allowable but expressing skepticism that such an imputation would be in the best interests of the children]; In re Marriage of LaBass & Munsee (1997) 56 Cal.App.4th 1331, 1339 [rejecting idea of an a priori bar to imputed income to supported parent].) There is no issue of imputed income to the payee parent here because the trial court imputed none and Husband does not challenge that lack-of-imputation in this appeal.
n9 There is no basis in law or fairness to impute income from undivided community assets solely to one spouse when the other spouse presumably has an ownership interest in those assets too. (Cf. In re Marriage of Schnabel (1993) 5 Cal.4th 704, 715 [spouse who was not employed by corporation had same right to inspect records of corporation as spouse who was so employed, where community owned stock in corporation].)
n10 The Dow, for example, hit its all-time high in January 2000, reaching 11723. At the end of 2001 it closed at 10021, down about 7 percent for the year. It was down another 17 percent for 2002, closing at 8341. Then it bounced back to 10453, or up about a quarter, in 2003. But even that was down about 10 percent from the 2000 high.
n11 We have to specify “this record” because this case concerns the order the trial court should have made in 2003, not the income of the parties at the time when, assuming the judgment of this court stands, the case reaches the trial court on remand in 2005. So there will be no need for any more evidence; it is simply a matter of properly calculating support from evidence adduced then. In fact, hopefully the parties will be able to agree on an income figure consistent with this opinion (we have already given a range of figures that constitute representative samples), run an Xspouse or a dissomaster themselves, and save themselves the expense of a hearing on remand.
n12 On the assumption that the parties do not settle the matter themselves, unless Judge Pollard is retired, disqualified pursuant to section 170.6, subdivision (a)(2) of the Code of Civil Procedure, or not available within the meaning of section 661 of the Code of Civil Procedure, the hearing on remand shall be assigned to her, regardless of the “panel” to which she may be currently assigned. (Abbott v. Mandiola (1999) 70 Cal.App.4th 676, 677-678 [judges who presided over proceeding on which another proceeding is based should hear the later proceeding].) In family law, it is particularly important that there be continuity. (Bidna v. Rosen (1993) 19 Cal.App.4th 27, 38-39.)
-END OF CASE-