11 N.J. Tax 482 | N.J. Tax Ct. | 1991
This case presents a question of the constitutionality of the use of gender-based mortality tables to value life estates and, thereby, determine transfer inheritance tax liability. The complaint was filed by the executor of the Estate of David M. Darrin after the New Jersey Transfer Inheritance Tax Bureau of the Division of Taxation in the Department of the Treasury (bureau) valued decedent’s widow’s life estate using a female mortality table. Taxpayer alleges that the bureau’s method of calculating tax liability constitutes invalid and invidious discrimination against women in violation of the United States and New Jersey Constitutions.
This court previously granted summary judgment in favor of taxpayer, holding, in part, that equal protection principles require the use of a gender-neutral mortality table in the valuation of life estates for inheritance tax purposes. Darrin Est. v. Taxation Div. Director, 9 N.J.Tax 419 (Tax Ct.1987).
As to the constitutionality of N.J.S.A. 54:36-2, concerning the use of the gender-based group averages to determine life expectancy, we are convinced that such an issue may have a significant impact not only on the particular statute involved, but on other statutes and proceedings in other areas of the law, including rules of court. Therefore, the issue should not be decided merely on affidavits of experts. The issue here “is a very important one involving highly significant policy considerations and obviously should not be decided” on less than a full record, after a plenary hearing with testimony by actuarians, statisticians or other experts in the discipline at issue. [232 N.J.Super. 437, 443, 557 A.2d 677 (App.Div.1989), app. dism. 118 N.J. 193, 570 A.2d 958 (1989); citation omitted]
A hearing was held to afford the parties an opportunity to present expert testimony on the use and accuracy of gender-neutral and gender-based statistical tables. Each party presented an expert report and the testimony of one actuary.
The facts as stipulated by the parties are set forth in the Tax Court and Appellate Division opinions and are summarized below.
David M. Darrin died testate on June 6, 1983, survived by his wife, Margaret A. Darrin, and three sons. His wife was 58 years old at the time of his death. His will created a marital trust, with income to be paid to his wife during her lifetime. The will also provided that the trustee could pay to her such principal as the trustee, in his discretion, felt advisable for her health, support, maintenance and best interests, and that upon his wife’s death the corpus of the trust would be distributed among his issue. All death taxes were to be paid by the executor from the residuary estate.
N.J.S.A. 54:36-2, effective July 1, 1978, directs that:
In determining the value of a life estate, annuity or estate for a term of years, the United States Life Tables, after December 81, 1970, Single Life Male 6% and Single Life Female 6%, published by the United States Department of Health, Education and Welfare ... shall be used and shall be effective with respect to estates of decedents dying on or after January 1, 1978.
The publication listed in the statute does not exist. The bureau therefore substituted publication 723A (12-70), entitled “Actuarial Values II: Factors at 6 Percent Involving One and Two Lives” (hereinafter referred to as publication 723A), a publica
In valuing Mrs. Darrin’s life estate, the Director applied the factor found in table A(2) of publication 723A, a life interest present worth table for females. For females age 58, the factor is .65988. For males age 58, the factor is .57778, which is found in table A(l) of publication 723A. Prior to 1978, the Director used gender-neutral tables in evaluating life estates. A gender-neutral factor for persons age 58 is .61860. Use of either the male factor or the gender-neutral factor would have resulted in a lower tax liability.
At trial, the actuary expert for taxpayer, Daniel M. Arnold, testified that the use of the female life interest table in publication 723A unfairly penalizes a majority of females because, during 1959-61, 80.6% of females had the same year of death as 80.6% of males. Thus, he stated, only 19.4% of the females lived longer than the males. He also noted that publication 723A is based on deaths in the three-year period 1959-1961 and United States life tables produced from the 1960 census, which tables have not been adjusted to account for changes since 1960.
I.
Equal Protection and the Standard of Review.
It is well established that equal protection guaranties are measured using a three-tier test, with the nature of the right or class at issue determining which tier is applicable. Where a fundamental right, or suspect class is at issue, a statute must satisfy the standard of strict scrutiny under which the statute will be found to be constitutional if the means chosen are necessary to promote a compelling governmental interest. When a semi-suspect class, such as gender, is involved, the classification must “ ‘serve important governmental objectives and must substantially relate to the achievement of those objectives.’ ” Greenberg v. Kimmelman, 99 N.J. 552, 565, 494 A.2d 294 (1985) (quoting Craig v. Boren, 429 U.S. 190, 197, 97 S.Ct. 451, 457, 50 L.Ed.2d 397 (1976)).
The initial question is whether the statute in fact discriminates based on gender. See Mississippi Univ. for Women v. Hogan, 458 U.S. 718, 723,102 S.Ct. 3331, 3335, 73 L.Ed.2d 1090 (1982). A statute that treats similarly-situated men and women differently is discriminatory. Manufacturers Hanover Trust Co. v. United States, 576 F.Supp. 837 (S.D.N.Y.1983), rev’d 775 F.2d 459, 463 (2 Cir.1985), cert. den. 475 U.S. 1095, 106 S.Ct. 1490, 89 L.Ed.2d 892 (1986). The use of gender-based mortality tables clearly discriminates against females because at virtually every age the factor for women is higher than for men, resulting in greater tax liability for female life tenants.
Thus, the intermediate standard of review for a semi-suspect class is appropriately applied to the subject case, and to prevail, the Director has the burden of proving that the classification based on gender is substantially related to the achievement of an important governmental objective.
II.
A. What Is the Important Governmental Objective?
I found in my prior opinion that accuracy in the valuation of life estates is the important governmental objective. The objective of the legislative change from gender-neutral to gender-based tables was to achieve increased accuracy in the life estate valuation process.
Prior to the 1978 change from a gender-neutral to gender-based mortality tables, the class under the statute was simply life tenants. With the change to gender-based mortality tables, two classes were created under the statute, female life tenants and male life tenants. For the purpose of equal protection analysis in this case, these are the two classes to be considered.
C. Is the Change in Classification Substantially Related to Achievement of an Important Governmental Objective?
Taxpayer makes three arguments to support its contention that the change to gender-based mortality tables is not substantially related to achievement of an important governmental objective: (1) increased accuracy in the life estate valuation process may be realized by considering other factors which impact life expectancy; (2) the approach to valuation used by the Director is based on insurance and annuity principles that use the law of large numbers, which are not relevant here; and (3) the increase in accuracy is not sufficient to justify the change from a gender-neutral table to gender-based tables.
(1) Other Factors Impact Life Expectancy.
Taxpayer’s expert states that race, health, smoking habits and family medical history all have a significant impact on determination of the life expectancy of an individual. This expert states that equity in taxation dictates that if these other significant factors are not to be considered, then all persons should be taxed by the same standard rather than differentiating based on gender alone.
Although there may be other ways to more fully achieve a statute’s objective, equal protection is not violated because the Legislature did not choose such other ways. See N.J. Chapt, Am. I.P. v. N.J. State Bd. of Prof. Planners, 48 N.J. 581, 600, 602, 227 A.2d 313 (1967). The Legislature determined to increase accuracy, and did so by changing from a gender-neutral table to gender-based tables. Whether it could have increased
(2) The Law of Large Numbers.
The law of large numbers states that when grouping individuals for the purpose of making predictions about those individuals, a sufficient number of events is needed to make predictability reliable. At the same time, the individuals within the grouping should share characteristics which significantly set them apart as a group from those not within the group.
For insurance and annuity purposes, the concept of risk transfer underscores the need to use the law of large numbers. Insurers must determine how much money has to be collected from plan participants to ensure that a sufficient sum of money is available when payment must be made to a participant.
Arnold agrees that the law of large numbers is applicable when predicting the average life expectancy of a group of individuals and that its use is necessary for insurance and annuity purposes. Arnold contends, however, that the reasons which support the use of the law of large numbers by insurers (e.g., the transfer of risk from individuals to a group) are absent in the valuation of individual life estates for transfer inheritance tax purposes.
The statute in question calls for valuation of the life estate. Taxpayer’s expert does not seriously contend that the law of large numbers is irrelevant in predicting the average life expectancy of a group. Rather, his testimony is directed to the applicability of the law of large numbers to an individual life tenant instead of to female life tenants as a class. The test for constitutional equal protection is directed to the class, not to individuals.
In their reports and testimony, both experts have referred to the two sources, publication 1252 and publication 723A. Publication 1252 uses, for each table therein, a hypothetical cohort
Publication 723A starts with the values from column 3 of the tables in publication 1252. These values are used to prepare the commutation tables at the selected discount rate of 6% (tables H(l) and (2)) and the present worth tables for a life estate and remainder (tables A(l) and (2)), all of which appear in publication 723A.
Arnold testified that the data from column 4 of tables 2 and 8 in publication 1252, when compared or “matched”, indicates that 80.6% of the females had the same year of death as 80.6% of the males. Lautzenheiser testified that the matching of deaths is a statistieally-irrelevant calculation in the projection of life. Arnold also testified that publication 723A is used by the Director to value life estates without adjustment for mortality trends after the period 1959-61. Lautzenheiser stated in her
The major disagreement between the experts concerns the matching death analysis. Briefly, the matching death analysis requires the determination of “overlapping” deaths for each age interval {e.g., at the end of an age interval, if the number of male deaths is ten and the number of female deaths is eight, the number of overlapping deaths is eight), the total of overlapping deaths for all age intervals and then the division of this total by the total number of people in the cohort (100,000). Arnold says that, given the high percentage of women who died at the same age as men, a minority of women outlived men and, therefore, use of a gender-based table which projects a longer life expectancy for all women does not sufficiently increase accuracy.
Lautzenheiser’s rejection of Arnold’s conclusions from the matching death analysis has four bases. The first begins with the definition of a life estate, which, she stated, includes “the probability that a person age x will survive to age x + t,” where t is a year or years. The probability of surviving to future years is not considered under the matching death analysis. It cannot be concluded that, at the age interval where matches occur, the probability of life expectancy for those who survive to the next age interval is the same for both genders, as “looking back” might indicate. Further, the matching death analysis looks only at deaths {i.e., those of the male and female cohorts who die at a particular age interval, without regard to those who survive to the next age interval) which, statistically speaking, is deficient for projecting the life expectancy of males as a group or females as a group.
The second basis refers to the “excess deaths” which are overlooked in the matching death analysis. The term “excess deaths” refers to the carryover to the next age interval of women who “survived” their equally-aged male counterparts {e.g., of the original ten, the two females who survived to the
The third basis addresses the calculation of probabilities. In the calculation of probabilities, equally-likely outcomes are grouped, and the probability is specific to that group. This expert stated that males and females with the same birth date are not equally likely to die at the same age because there are biological differences between males and females.
The fourth basis goes to the degree of accuracy. Assuming that 80.6% of the deaths of males and females matched, Lautzenheiser contends that it cannot be inferred that the use of the gender-neutral table would result in more accuracy than the use of gender-based tables for the other 19.4%. Further, Lautzenheiser contends that the use of gender-based tables provides greater accuracy for 100% of the class, exact accuracy for 80.6% of the class and more accuracy for the remaining 19.4%, an increase in overall accuracy which is statistically significant. In contrast, this expert stated, when the results from use of the gender-neutral table are compared with the results from the use of either gender-based table, the difference in forecasted life expectancy for any age interval indicates a significant inaccuracy in the forecasting of life expectancy of either gender when the gender-neutral table is used.
III.
As a result of the hearing, I find that for the purpose of the statutory requirement to determine the value of the life estate, the probability of living into the future is a critical consideration. The matching death analysis has been offered for the purpose of showing that the use of gender-based tables does not result in significantly greater accuracy in forecasting life expectancy than does the use of a gender-neutral table. However, the matching death analysis is flawed because it does not consider those who survive to the next age interval and, therefore, fails to consider the probability of life expectancy. It is further flawed because when it compares male and female deaths it does so without consideration of the different mortality rates which were used to calculate those deaths. Thus, the matching death analysis is not reliable support for taxpayer’s position.
Further, as a result of the hearing, I find that, in forecasting life expectancy, the degree of accuracy associated with the use of gender-based tables for males as a class and females as a class is greater than that associated with the use of the gender-neutral table, and taxpayer’s expert has conceded this greater degree of accuracy. In my prior opinion, I found that the 80.6% overlap evidenced in the matching death analysis indicated that the use of the gender-based table resulted in less accuracy in forecasting the life expectancy for the 80.6%. The reasoning in the prior opinion was based on looking back at the results of deaths, which the matching death analysis does. I find from the expert testimony that, in forecasting life expectancy, looking back at deaths alone is insufficient. Therefore, I find that in forecasting life expectancy, those who survive to the next age interval relative to those who do not must also be considered. Because the matching death analysis does not consider these survivors, it is insufficient to negate a finding that the use of gender-based tables in forecasting life expectancy results in increased accuracy. Further, the Director’s expert
As to the use of tables which do not reflect the most current mortality experience, this argument by taxpayer comes under the heading of better ways to increase accuracy, and I have found such considerations to be irrelevant to this analysis.
I therefore find, based on the above reasons, that the Director has met the burden of showing a substantial relationship between the use of gender-based mortality tables in the forecasting of life expectancy and achievement of the important governmental objective of increased accuracy in the valuation of life estates.
In Manufacturers Hanover Trust Company v. United States, supra, the court stated:
While there is no mechanical test for determining whether or not a practice satisfies the substantial relationship test, case law does make it clear that at least four particular matters must be explored and weighed: (1) aggregate impact on class; (2) demeaning generalization; (3) stereotyped assumptions; and (4) flawed use of statistics. [775 F.2d at 465].
Although I note that a disparate impact on female life tenants may result from the use of gender-based mortality tables, there is no evidence that the other three factors enunciated by the court therein exist here. Given the four factors and the weight I assign to each, I find that any disparate impact here is insufficient to overcome my finding that a substantial relationship exists between the use of gender-based tables and increased accuracy in the valuation of life estates.
Conclusion.
In summary, I find that no equal protection violation exists where the Legislature, in pursuance of increased accuracy in the life estate valuation process, has chosen to use gender-based mortality tables for inheritance tax purposes. Counsel for Director will submit an appropriate order under the five-day rule.
Prior to the decision on the motion, the Director stated, "[d]ue to the relatively high cost of the expert testimony and the fiscal constraints of the Division of Taxation, the Director will not be submitting any rebuttal testimony.”
It has been indicated to the court that a comparable table for a period after December 31, 1970 was not available, and neither party has urged the adoption of a table based on 1970 or later census information. Taxpayer challenges the use of data from the period 1959-1961 not for the purpose of seeking more current data but to support its contention that the Director abused his authority by using gender-based tables. The change to a more current gender-based table would not affect taxpayer's underlying equal protection argument that a gender-neutral table, not a gender-based table, be used. Presumably, more current life tables would have longer life expectancies and therefore would result in a higher value for the subject life estate.
The law of large numbers refers to the principle that by recording the repeating of an event a large number of times, you can most accurately predict the probability of the event occurring in the future. According to Lautzenheiser, mortality tables are based on this principle.
I note that the gender discrimination standard is different here than in Title VII cases where the statutory language groups sex with race, color, religion and nationality in prohibiting employment discrimination against any individual. See Arizona Governing Committee for Tax Deferred Annuity and Deferred Compensation Plan v. Norris, 463 U.S. 1073, 103 S.Ct. 3492, 77 L.Ed.2d 1236 (1983); Los Angeles Dep’t. of Water and Power v. Manhart, 435 U.S. 702, 98 S.Ct. 1370, 55 L.Ed.2d 657 (1978).
For example, in predicting the probability of rolling a seven with dice, all combinations of seven are included, but all combinations which total any other number are excluded.
For actuarial purposes, a cohort is the group of individuals under study.
Column 2 of tables 2 (males) and 3 (females) reflect actual mortality rates during the period 1959-61. Referring to table 2, the hypothetical cohort begins with 100,000 bora alive, thus the value "100,000" in column 3 for the age interval "Years 0-1”. The number “2913” in column 4 is determined by multiplying 100,000 by the mortality rate ".02913” for the age interval "0-1”. For the age interval “1-2”, the "100,000" is reduced by the "2913", resulting in the number “97087” in column 3 for the age interval "1-2”. The number “176” in column 4 for the age interval "1-2” is derived by multiplying the “97087” by the mortality rate ".00181" for the age interval “1-2”.
In support of this expert’s contention, she points to the different chromosome makeup of males (which makeup, she states, renders males less able to adjust for a defective gene), studies which indicate a higher perinatal mortality rate for males, and an advantage by females over males in resisting or overcoming diseases, particularly inherited diseases.
By way of example, Director’s expert explains that the matching of gender-based data is similar to matching the results of rolling dice and concluding that there is no difference between the probability of rolling a seven and the probability of rolling a five.