Order, Family Court, New York County (George Jurow, J.), entered November 19, 1990, which, inter alia, denied petitioner’s objections to the order of the Hearing Examiner, dated September 11, 1990, which calculated respondent’s [Raymond S.] support obligation for five children living in two separate households utilizing the statutory percentage for five children (35%) and dividing it equally among them, unanimously reversed, on the law, without costs, and the matter remanded for calculation of a new child support award consistent with this decision and order.
Order, Family Court, New York County (George Jurow, J.), entered October 31, 1990, which, inter alia, denied petitioner’s objections to the order of the Hearing Examiner, dated August 13, 1990, which calculated respondent’s [Manuel S.] support *511obligation for four children, who lived in three separate households, by utilizing the statutory percentage for four children (31%) and dividing it equally among them, unanimously reversed, on the law, without costs, and the matter remanded for calculation of a new child support award consistent with this decision and order.
Order, Family Court, New York County (George Jurow, J.), entered October 31, 1990, which, inter alia, denied petitioner’s objections to the order of the Hearing Examiner, dated August 7, 1990, which calculated respondent’s [Daniel T.] support obligation for four children, who lived in two separate households, by utilizing the statutory percentage for four children (31%) and dividing it equally among them, unanimously reversed, on the law, without costs, and the matter remanded for calculation of a new child support award consistent with this decision and order.
The three cases consolidated for appeal before this court present the question whether it is proper, in calculating a noncustodial parent’s child support obligation for children in more than one household, to consolidate the cases, look only to the statutorily mandated support percentage for the number of children and divide that amount equally among the children, even though they have different custodial parents and reside in separate households.
In the Raymond S. action, initial child support was ordered on behalf of one mother and her three children, taking into account support the father (Raymond S.) was paying pursuant to a four year old order of child support for two of his other children from his marriage to a different woman. Raymond filed objections to the new order and moved for downward modification of the prior order. The Family Court (Kaplan, J.) ordered the consolidation of both actions and instructed the Hearing Examiner to calculate Raymond’s support obligation for the five children who lived in two separate households, by utilizing the statutory percentage for five children (35%) and dividing it equally among them.
In the Manuel S. action, three initial child support orders were sought against the same noncustodial father (Manuel S.) on behalf of three custodial mothers and their four children. Manuel filed objections to orders made by the Hearing Examiners and also petitioned for a downward modification of each prior order, based upon the additional obligation placed on him by each subsequent order. The Family Court (Kaplan, J.) ordered the consolidation of all three cases and instructed the Hearing Examiner to calculate Manuel’s child support obliga*512tion for the four children, who lived in three separate households, by utilizing the statutory percentage for four children (31%) and dividing it equally among them.
In the Daniel T. action, initial child support was ordered on behalf of one mother and her two children, taking into account support the father (Daniel T.) was paying pursuant to a three year old order of child support for two of his other children, born to a different woman. Daniel filed objections to the new order and moved for downward modification of the prior order. The Family Court (Kaplan, J.) ordered the consolidation of both actions and instructed the Hearing Examiner to calculate Daniel’s child support obligation for the four children who lived in two separate households, by utilizing the statutory percentage for four children (31%) and dividing it equally among them. For reasons that follow, we conclude that the method of computation in each of these cases was improper, and the orders resulting therefrom must be reversed.
The Child Support Standards Act (CSSA; L 1989, ch 567) which became effective September 15, 1989, amended, inter alia, Family Court Act § 413 and Domestic Relations Law §§236 (B) (7) and 240 to establish a presumptive numerical formula to the first $80,000 of "combined parental income” when calculating child support awards in divorce, custody, paternity and child support proceedings. Under the CSSA formula, the gross or total income for each of the child’s parents is first computed pursuant to Family Court Act § 413 (1) (b) (5). Deductions are then made for various items in arriving at what is commonly referred to as the parent’s adjusted gross income, but which is termed in the statute as simply "income”. (Id.) One of those allowed deductions is child support that is "actually paid pursuant to court order or written agreement” for a child not the subject of the pending action. (Family Ct Act § 413 [1] [b] [5] [vii] [D].)
The foregoing "income” figures for each parent are then added together. (Family Ct Act §413 [1] [c] [1].) The first $80,000 of that "combined parental income” is thereafter multiplied by the applicable child support percentage set forth in section 413 (1) (b) (3), with each parent’s share of the resulting "basic child support obligation” determined by dividing that sum between the parents in the same proportion as each parent’s income bears to the "combined parental income”. (Family Ct Act § 413 [1] [c] [2].)
The statutory percentage for one child is 17%. If there is *513more than one child, the applicable percentage has not been established simply by multiplication of the 17% figure by the number of children. Rather, the percentages are 25, 29 and 31%, respectively, for two, three and four children; and "no less than” 35% for five or more children. This is because the percentages are to be applied on a per household basis, with the controlling percentage for each such home determined according to how many children are living with the same custodial parent, thereby requiring lower per capita child-rearing costs due to the benefit of shared resources. (For a detailed discussion of the CSSA and how support awards are computed under it, see, Reichler and Lefcourt, New Legislation: The Child-Support Standards Act, NYLJ, June 30, 1989, at 1, col 1.)
That this was the Legislature’s intent is evident from our examination of the legislative history of the CSSA, and from a letter to the New York Law Journal (Aug. 8, 1990, at 2, col 6) entitled Explaining Rationale for Support Standards, written by Helene E. Weinstein, one of the principal co-sponsors of the CSSA legislation in the State Assembly. In that letter Ms. Weinstein criticized a decision written by Judge Kaplan, Commissioner of Social Servs. v Jose E. (NYLJ, July 18, 1990, at 20, col 6 [Fam Ct, NY County]). That case involved two families who were seeking child support from the same father, in two separate cases. There were two children in one household, and one in the other. Judge Kaplan had taken the 29% figure for three children from Family Court Act § 413 (1) (b) (3) (iii), multiplied that percentage against the combined parental income, and divided by three, even though the three children lived in two separate households. (We note that in the three cases now before us, Judge Kaplan had ordered the Hearing Examiners to compute child support in accordance with his decision in Commissioner of Social Servs. v Jose E., supra).* With respect to Judge Kaplan’s computation in that case, Ms. Weinstein stated:
"In such cases, dividing 29 percent between the households would result in an artificially low total amount of child support. While it may be infrequent that the court will have two families before it at the same time, it is important to *514understand a basic principle of the CSSA so that unfairness will not result.
"The percentages in the CSSA are based upon the proportion of income required to support a child, or children, in a household. The numbers reflect the effect of economy of scale, which works like this: If one child requires 17 percent of income, two children in the same household do not require 34 percent. This is because there are duplicated costs, and the cost per child is less when two or more children live in the same household. Hence, the figure of 25 percent for two children, rather than 34 percent.
"Conversely, if two children [in the same household] require 25 percent of income, one child requires more than 12 Vi percent because the economy of scale works only when there are two or more children. Hence, the figure of 17 percent for one child, rather than 12 Vi percent.
"Once this principle is understood, it is apparent that where the children live in more than one household, it would never be appropriate to look at the percentage for that number of children and divide it among the households.” (Op. cit.)
There are only two statutorily-permitted exceptions from the basic CSSA formula. One is where the computed "basic child support obligation” reduces the non-custodial parent’s income below a "self-support reserve”, which is defined in section 413 (1) (b) (6) as 135% of the "poverty income guidelines amount for a single person”. Even in that situation, a minimum support obligation is still mandated. (Family Ct Act § 413 [1] [d], [g]; see, Matter of Beaudoin v Joseph K., 165 AD2d 359.) The other exception is where the court determines that the non-custodial parent’s pro rata share of the basic child support obligation is "unjust or inappropriate” based on consideration of the ten factors set forth in Family Court Act § 413 (1) (f). Whenever the court varies from the statutory formula under the "unjust or inappropriate” exception, it must set forth, in a written order, the factors it considered and the reasons for the level of support ordered. (Family Ct Act § 413 [1] [g].)
We emphasize that the percentages set forth in Family Court Act § 413 (1) (b) (3) were not enacted as a tax on the parents’ income, but rather were based on studies indicating what portion of parents’ combined income is equitably apportioned to child support when the children reside in a single household. "The needs of the children are paramount, regardless of the marital status of the parents.” (Gelb v Brown, 163 *515AD2d 189, 191.) The method of calculation ordered by Judge Kaplan in each of the actions herein resulted in child support orders lower than those contemplated by the Legislature when the CSSA was enacted, and those orders must accordingly be reversed and the matters remanded for calculations of new child support awards consistent with this decision and order. Concur—Sullivan, J. P., Carro, Milonas, Asch and Rubin, JJ.
Judge Kaplan has since rejected his interpretation of the CSSA as reflected in the cases presented herein, and in subsequent cases has construed the child support percentage to correspond to the number of children living in the household that is the subject of the proceeding, rather than the total number of children fathered by a particular man, but living in separate households.