See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/387718326

Sexual Freedom, Fertility, and Time Preferences: Panel Data Analysis From 1960
to 2010

Article in The American Economist · January 2025

DOI: 10.1177/05694345241311552

CITATIONS READS
0 22

2 authors:

Feler Bose Jeffry Jacob

Indiana University East Bethel University
55 PUBLICATIONS 161 CITATIONS 13 PUBLICATIONS 51 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Feler Bose on 08 January 2025.

The user has requested enhancement of the downloaded file.

 Sexual Freedom, Fertility, and Time Preferences: Panel Data Analysis from 1960 to 2010

Feler Bose*
School of Business and Economics
Indiana University, East
Richmond, IN 47374, USA
Email: [email protected]

Jeffry Jacob
Department of Business and Economics,
Bethel University,
St. Paul, MN 55112, USA
Email: [email protected]

This is the Accepted Manuscript

Citation:
Bose, Feler and Jeffry Jacob, “Sexual Freedom, Fertility, and Time Preferences: Panel Data
Analysis From 1960 to 2010,” The American Economist, 2025, Forthcoming.
https://doi.org/10.1177/05694345241311552

Key Words: Sexual Freedom Index; Panel Data Analysis; Fertility; Time Preferences; Citizen
Ideology; Sorokin Thesis;
JEL Code: K36, K42, K49
*corresponding author.

Disclosures:
-The authors have no relevant financial or non-financial interests to disclose.
-The authors did not receive support from any organization for the submitted work.
-This article does not contain any studies with human participants or animals performed by any of the
authors.
- The data supporting this study's findings are available from the corresponding author, FB, upon
reasonable request.
--All authors have equally contributed to this paper.

1
 Sexual Freedom, Fertility, and Time Preferences: Panel Data Analysis from 1960 to 2010

Abstract:
Standard economic theory offers a demand-and-supply view of fertility. With increasing
economic development, higher income levels, dropping mortality rates, and increases in
education levels resulting in increased female labor force participation, there is a decreased
demand for children. These determinants of fertility could also be seen as ‘backward-looking’
determinants since they deal with variables that have already happened to the person. Another
theory for explaining decreased fertility could be the time preference theory, which is a ‘forward-
looking’ variable based on a person’s future expectations (Bellani et al., 2021). Sexual freedom
is seen as a measure of time preference (Banfield 1974) and is expected to affect fertility
(Sorokin, 1956). For this reason, we will study the impact of increasing sexual freedom on
fertility, controlling for various state-level characteristics. To control for endogeneity between
fertility and other explanatory variables, we will use fixed effects instrumental variable
estimation.

2
 1. Introduction
Standard economic theory offers a demand-and-supply view of fertility. With increasing

economic development, higher income levels, dropping mortality rates, and increases in

education levels resulting in increased female labor force participation, there is a decreased

demand for children. This results in a reduced fertility rate. These determinants of fertility could

also be seen as ‘backward-looking’ determinants since they deal with variables that have already

happened to the person. Another theory for explaining decreased fertility could be the theory of

time preference, which is a ‘forward-looking’ variable and is based on a person’s future

expectations (Bellani et al., 2021).

Several backward-looking determinants in the fertility literature include the relationship

between income or wages and fertility (Jones et al., 2010), the relationship between the fertility

of successive generations (Murphy & Wang, 2001), the relationship between the later onset of

adult activities and its implications for fertility and also economic opportunities (Billari &

Tabellini, 2010), the impact of having children vs. being childless on having future children

(Dommermuth et al., 2015), the relationship between status (education, class, rank, etc.) on

fertility (Skirbekk, 2008), the impact of different types of debt levels on fertility (Nau et al.,

2015), among others.

Further, the naturalist theories of fertility come into three categories. The first category

theorizes that reduced fertility is because of the change in the environment, i.e., our modern

industrial environment is different than the primitive environment of our forebears. The second

theory is a substitution theory that in the current environment parents invest more in their

children and hence they produce fewer kids, a tradeoff between quantity and quality. The final

theory is focused on cultural evolution resulting from an arms race. When influential people

3
adopt a trait, others rush to copy it. In this case, getting and valuing more education results in the

reduction of mortality (increased survivability) and fertility, which then gets transmitted to others

who copy it (Wodarz et al., 2020).

Another line of literature on fertility, sees fertility decisions as dynamic. There is a

momentum built in fertility as past fertility affects present fertility. On the other hand, fertility

decisions are also forward-looking since a decision to have a child impacts many future

decisions. Hence time preferences play a role in fertility. Time Preference is a psychological

concept that suggests that people, when all other factors are equal, generally prefer to achieve a

particular outcome sooner rather than later. This preference can be seen as a reflection of an

individual’s patience and their ability to delay gratification.

People with High Time Preferences (high discount rates) are often characterized as those

who live in the moment. They are typically impulsive and impatient, preferring immediate

satisfaction over delayed rewards (Peters & Büchel, 2011). On the other hand, individuals with

Low Time Preferences (low discount rates) are those who can delay gratification and plan for the

future. They are often seen as patient and strategic, willing to forgo immediate satisfaction in

favor of long-term gains (Mulligan, 2008, p. 69).

When tying time preferences to fertility, a person with high time preference might make

hazardous choices, choose a poor partner, and hence might be viewed by their partners as

inadequate parents and hence not have children. Further, having children would be seen as

having immediate costs both in terms of time and money, hence discouraging having children

(Folbre, 2008). Low time preferences individuals take more time to search for a life partner,

seek a good job, then seek to have children. Further, they will be concerned about long-term

costs and benefits instead of short-term net benefits. These long-term benefits could be for

4
example, a child’s support to one’s parents as one gets older (Dykstra & Fokkema, 2011). This

line of reasoning would suggest that higher time preference would mean less fertility and lower

time preference individuals have higher fertility.

On the contrary, if children provide immediate value by increasing relationship stability

and attachment (Hill, 1988), high time preference individuals might desire to have children right

away resulting in for example teenage pregnancy (Golsteyn et al., 2014). Those with low time

preferences could wait for the perfect job, the perfect partner, and procrastinate having children

resulting in the inability to have children due to the reproductive opportunity window closing.

Additionally, their long-term calculation with awareness of the increasing costs of having

children (Craig et al., 2014) might hinder them from having children. This would suggest that

higher time preference means more fertility and lower time preference individuals have lower

fertility. Because of these two possible pathways, the relationship between time preferences and

fertility might be nonlinear (Bellani et al., 2021).

Pitirim Sorokin, the founder of the sociology department at Harvard University, in his

book “The American Sex Revolution” has suggested that societies that increase sexual freedom

or “sexual anarchy” experience various social ills including lower fertility rates. He finds

support for his hypothesis after surveying several cultures (e.g., Rome and Greece) and royal

families. “As a rule, communities preoccupied by the hunt for promiscuous sex pleasure are

little, if at all, about having children, for they are obstacles to the full enjoyment of libertinism.

This viewpoint prompts sex devotees to resort to contraceptives, abortions, and other means for

preventing childbirth” (Sorokin, 1956, pp. 78-79). Even if people in these cultures marry, their

marriages are either childless or produce one or two children resulting in the “voluntary suicide”

of the society.

5
 In this paper, we introduce a new index of sexual freedom. This unique dataset tracks

laws that allow for sexual relationships beyond that of a man and woman in marriage as the only

legal sexual arrangement. The data spans 51 years from 1960 to 2010 for all 50 states of the

United States (U.S.). Historically, in the U.S. both family law and criminal law have supported

limiting sexual activity to the realm of marriage (Murray, 2009) but this has changed over the

last few decades. Our goal in this paper is to study the impact of sexual freedom on fertility rates

(Sorokin’s thesis) using the novel dataset and show the link with the time preference literature on

fertility.

Section 2 discusses the issue of time preferences and their relation to sexual freedom.

Section 3 details the data collection process, section 4 provides our empirical strategy, section 5

discusses the results, and section 6 concludes.

2. Literature Review of Time Preference and Sexual Freedom

Time preference plays an important role in human action including as stated earlier on

fertility decisions. Various factors affect people’s time preferences. They include “external,

biological, personal, and social or institutional ones” (Hoppe, 2001, p. 3). Biological factors

include just the aging process. When one is a child, one will have high time preferences, and this

tends to decline as one ages and increases as one nears death. External factors affecting time

preference include “property-right security, physical security generally, and observed behavior of

others in the community” (Mulligan, 2008, p. 81).

Religion also affects time preferences. Max Weber noted Calvinism encourages low

time preferences. “When the limitation of consumption is combined with this release of

acquisitive activity, the inevitable practical result is obvious: accumulation of capital through

6
ascetic compulsion to save. The restraints which were imposed upon the consumption of wealth

naturally served to increase it by making possible the productive investment of capital… the

greater simplicity of life in the more seriously religious circles, in combination with great wealth,

led to an excessive propensity to accumulations.” (1930 p. 85). Further, some religious

groups/sects like messianic and millenarian sects can encourage high time preferences as they

wait for the end of the world (Mulligan, 2008, p. 83).

Time preferences affect social and economic outcomes. Even measuring discount rates

of individuals when they are teenagers can be used to predict the future in terms of “school

performance, health, labour supply and lifetime income” (Golsteyn et al., 2014). The further in

the future one goes, the discrepancy caused by different discount rates is more obvious. The

differences in time preference seems to be mediated by early human capital investment.

Moreover, those with high discount rates (high time preference) correlate with sexual behaviors

and outcomes related to sexual health. These include instances of engaging in sexual activity, the

onset of sexual activity at an earlier age, and teenage pregnancy (Chesson et al., 2006). Even at

the macro-level, a paper finds that countries that exhibit more patience (more future-oriented) are

wealthier and more developed (Sunde et al., 2021).

Time preferences are related to the issue of self-control. Hence the lack of self-control

“can manifest itself in many ways, including elevated emotional lability [sic], tendency to fly off

the handle, low tolerance for frustration, lack of persistence, short attention span, high

distractibility, frequent shifting from activity to activity, being restless, being overactive, acting

before thinking, having problems waiting for something good, and having difficulty taking turns”

which impacts one's actions in the sexual marketplace resulting in unsafe sex (Belsky et al.,

2020).

7
 In the sexual marketplace, the notion of time preference also applies. Banfield writes

that the “lower classes” (those with high time preferences) have no sense of the future. A higher

time preference individual considers his “bodily needs (especially for sex)…take precedence

over everything else—certainly over any work routine” (1974, p. 61). Further, future-oriented

(lower time preference) cultures teach “the individual that he would be cheating himself if he

allowed gratification of his impulses (for example, for sex or violence) to interfere with his

provision for the future” (Banfield, 1974, p. 57).

Some empirical work regarding sexual freedom has been completed. One study has

found that when men see pretty women, they discount the future more as it has to do with mating

opportunities (Wilson & Daly, 2004). However, it is not clear whether showing cues of pretty

women changes underlying time preferences, i.e., do men want to copulate now or are they

willing to wait till say marriage to consummate the relationship? Two Papers (Bose, 2015; Bose

& Jacob, 2018) use the Sexual Freedom Index for a shorter time frame of 21 years to determine

the factors that determine sexual freedom and one key finding is that citizen ideology has a

statistically significant impact on the Sexual Freedom Index.

The psychology literature shows that citizen ideology is tied to time preferences. Right

ideology shows greater self-control than the left i.e., those on the right have lower time

preferences. These seem to be mediated by the role freewill beliefs play for the individual.

However, in some instances, those on the left can use external factors to display self-control and

achieve results (Clarkson et al., 2015). Left-wing political ideology is also a predictor in crime,

which indicates high time preference. This could be due to the left’s openness to new and

exciting experiences. In contrast, those on the right prefer law and order and respect for authority

(Wright et al., 2017). Adventurous and risky sexual activity, including the use of toys and one-

8
night stands, is strongly correlated with left-wing political ideology, while right-leaning ideology

tends to report engaging in a narrower and more traditional scope of sexual activity (missionary

position and romantic kissing) (Hatemi et al., 2017). This ties in with Sorokin’s thesis, that

sexually free societies will be made up of high time preference individuals, which will then

affect fertility decisions.

Bose and Van Duyn (2020) analyze party platforms for 22 OECD countries and find that

right parties tended to emphasize more economic freedom vs. sexual freedom when compared to

left parties. They hypothesize that these differences between the parties are because the left

parties attract people with higher time preferences whereas the right parties attract individuals

with lower time preferences.

Becker and Mulligan also try to understand how people might be able to change their

time preferences. They suggest that an individual can change their time preferences by spending

time and effort appreciating future pleasures (1997, p. 734). Education is one method of

lowering time preferences as “through repeated practice at problem-solving, schooling helps

children learn the art of scenario simulation. Thus, educated people should be more productive at

reducing the remoteness of future pleasures” (Becker & Mulligan, 1997).

In this paper, we study the impact of increasing sexual freedom and citizen ideology on

fertility in the United States using state-level panel data from 1950-2010. We will use fixed

effects instrumental variable estimation to control for endogeneity between fertility and other

explanatory variables.

To our knowledge, this is the first study investigating Sorokin’s thesis, in which we

investigate the impact of sexual freedom on fertility in the U.S. at the state level.

9
 3. Data
Our variables have been collected from 1960 to 2010 at the state level for all 50 states.

One of the challenges in historical data collection is dealing with missing data and ensuring the

quality of the available data. We addressed these issues as best as possible, but challenges

continue to remain.

3.a. Independent Variables

The sexual freedom index codes fifteen variables (laws): bestiality, abortion, sodomy,

fornication, pimp laws, prostitution laws, age of consent, adultery, homosexual marriage and its

recognition, polygamy, divorce, cohabitation, miscegenation, and marrying close blood. Each

variable/law was coded on a four-point scale (0, 1, 2, 3). A value of 0 reflects the lowest

possible amount of freedom or the most regulation, and a three represents the greatest amount of

freedom or no regulation. We use a similar coding method as found in Bose’s paper (2021).

The citizen ideology measure comes from Richard Fording’s website1. The citizen

ideology is the ‘revised 1960-2010 citizen ideology series’ (Berry et al., 1998). The larger the

number, the more leftist/liberal the active electorate of the state. We did not use the state

government ideology measure of Berry et al., (2010) as it measures the ideal points of the state’s

congressional delegation and not the state government directly. Instead, we use the modified

Ranney Index. The modified Ranney index is an index that averages percent Democrats in the

House and Senate, and the percentage of votes for the Democrat’s gubernatorial candidate,

accounting further for whether they have control of all three institutions of governance (house,

senate, and governor’s office) in the state. The Ranney Index formula is below. We obtained the

1
http://rcfording.wordpress.com/state-ideology-data/ (accessed June 1, 2013).

10
data to create the Ranney Index from Klarner (2013) and used current-year information instead

of running averages over many years.

Ranney Index = [(Percent Vote for Democrat’s Candidate for Governor) + (Percent
Democrats in House) + (Percent Democrats in Senate) + (Control Variable)] / 4 (1)
Control Variable: If democrats control house, senate, and governorship: = 1 (100%),
Otherwise = 0 (0%). If democrats have control of only some branches: = .5 (50%)

A Ranney Index closer to 1 indicates full Democrat control of the state, whereas, a

number closer to 0 indicates Republican control. Per the party platform literature, full Democrat

control indicates that the voters in the state have high time preferences, as the party platform of

the state Democrat party would be focused on sexual liberty and other high time preference

lifestyles (Bose & Van Duyn, 2020).

Other state variables include real gross state product per capita and shares of Catholics,

and protestants in the state’s population. This data comes from the Association of Religion Data

Archives2. We also include the share of whites in the total population to control for the effect of

race on fertility. Tables 1 and 2 show the summary statistics and the correlation coefficients

respectively.

Table 1: Summary Statistics

Variable Obs Mean Std. Dev. Min Max
General Fertility Rate 351 66.20 8.12 50.08 113.16
Sexual Freedom Index 550 1.38 0.37 0.54 2.31
Citizen Ideology 441 46.90 15.71 4.75 90.17
Ranney Index 539 0.56 0.18 0.14 1.00
Log GSP per capita 550 9.59 0.94 7.60 11.15
% of catholics 546 0.19 0.13 0.01 0.64
% of protestants 546 0.12 0.07 0.01 0.37
% of whites 550 0.82 0.14 0.23 1.00

2
Thearda.com (accessed October 1, 2013).

11
 Table 2: Correlation Coefficients
Log
General Sexual GSP
Fertility Freedom Citizen Ranney per % of % of % of
Rate Index Ideology Index capita Catholics Protestants whites
General Fertility Rate 1
Sexual Freedom Index -0.014 1.000
Citizen Ideology -0.429 0.215 1.000
Ranney Index -0.141 -0.066 0.174 1.000
Log GSP per capita -0.003 0.209 0.047 -0.175 1.000
% of catholic -0.265 0.109 0.630 -0.042 0.068 1.000
% of protestants -0.218 -0.041 0.085 -0.143 -0.263 -0.015 1.000
% of whites -0.200 -0.171 0.015 -0.399 -0.319 0.106 0.390 1.000

3.b. Dependent Variable

Measuring fertility was a challenge. There are three different measures of fertility

(Lucas, 2018; Mosley, 2006). The crude birth rate is the only data that goes back to 1960. The

crude birth rate is an estimate of fertility defined as the number of live births per 1000 total

population in a given year. It is not good for comparing fertility across populations, as variations

in age distribution will affect the birth rate

𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝐵𝑖𝑟𝑡ℎ𝑠/𝑌𝑒𝑎𝑟
Crude Birth Rate=. × 1000
𝑇𝑜𝑡𝑎𝑙 𝑀𝑖𝑑−𝑦𝑒𝑎𝑟 𝑃𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛

Only a “crude” estimate of fertility, all of the population included in the denominator is

not exposed to the risk of pregnancy

The general fertility rate is of higher quality measure as it is the number of live births per

1000 women ages 15-49 within a given year. It is approximately 4 times the crude birth rate. It

also allows for better comparison across populations. The data for the general fertility rate was

available from 1980.

𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝐵𝑖𝑟𝑡ℎ𝑠/𝑌𝑒𝑎𝑟
General Fertility Rate= 𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑤𝑜𝑚𝑒𝑛 𝑎𝑔𝑒𝑠 15 𝑡𝑜 49 × 1000

12
 The total fertility rate is the best single measure of fertility across populations. It is the

expected births to a woman by the time she ended childbearing if she were to pass through all her

childbearing years conforming to the age-specific fertility rates of a given year. Five is the width

of the age group interval. This data was available from 1990. All variables are averaged over

five-year periods.

4. Model
In this paper, we study the effect of the sexual freedom index, citizen ideology, and

government ideology (as measured by the Ranney Index) on fertility across the fifty states in the

US. We begin our analysis by examining this issue using static panel data models.

Our estimating equation is

gfrit = 0 + 1sfiit + 2 X1it + 3 X 2it + i + i,t (1)

where gfrit is the general fertility rate for state i at time period t, sfiit is the index of state i’s

sexual freedom, X 1it is the set of state ideology variables and X 2 it is the set of other state

control variables  i is a state-specific fixed effect, and  it is the idiosyncratic error term. We

begin by reporting random and fixed effects models. One potential problem in the estimation of

(1) is that several of the right-side variables could be endogenous. For example, not only could

ideology and SFI variables affect fertility, but be in turn determined by it. This is especially true

in a panel data context where people may move in and out of states based on their attitudes

toward these issues. Typically, endogeneity in the model can be addressed by finding external

instruments. Finding appropriate instruments for the endogenous variables in our context is very

difficult. One alternative to external instruments is to employ internal instruments from the

model. One such approach is suggested by Lewbel (1997) which exploits the curvature of the

13
variable by using as instruments different orders of centered moments of the endogenous

variable (i.e., the deviation of the variable from its global mean, raised to the power two, three,

etc.). We use this approach in our FEIV model.

As a robustness check, we also estimate our model using a dynamic panel data approach

which allows for general fertility rates to be influenced by past periods' values. The estimating

equation is:

gfrit = 0 +  gfrit −1 + 1sfiit + 2 X1it + 3 X 2it + i + i,t (2)

Where gfrit −1 is the past period’s fertility rate. The problem of endogeneity of the

explanatory variables can easily be addressed in a dynamic panel data context as well, where

higher-order lags of the endogenous variables can be used as instruments.

The standard approach is to estimate the models like the ones in equation 2 in the first

differences thus eliminating the individual fixed effects. Specifically, the estimating model thus

becomes:

yi,t − yi,t −1 = 1 ( yi,t −1 − yi,t −2 ) + 2 ( X it − X i,t −1 ) + (i,t − i,t −1 ) (3)

The problem in estimating Eq. (3) is the endogeneity introduced by the lag-dependent variable

E[( yi,t −1 − yi ,t −2 )(i ,t − i ,t −1 )]  0

since . The dynamic panel data estimator developed by Holtz-

Eakin et al., (1988) and Arellano and Bond (1991) addresses this issue by using two periods or

more lags of the dependent variable as instruments for the differenced lagged dependent variable

since

E[ yi,t −s (i,t − i,t −1 )] = 0, for t = 3, 4...T and s  2

.

14
Blundell and Bond (1998) show that this difference estimator may not perform well when there

is persistence in the lagged dependent variable and demonstrate at the systems GMM estimator,

initially proposed by Arellano and Bover (1995), may be better suited. The systems GMM

estimator is based on the idea that additional moment conditions can be introduced by adding a

level equation to the differenced equation and using lagged differences of the explanatory

variables as instruments for the level equation since E[i,t ( yi,t −1 − yi,t −2 )] = 0, for t = 3, 4...T .

To avoid the problem of instrument proliferation, which can overfit the model, we use a

collapsed instrument set as developed by Roodman (2009). Two-step dynamic panel data models

can result in inefficient standard errors in finite samples. We report bias-corrected robust

standard errors to avoid this problem (Windmeijer, 2005).

5. Analysis
Table 3 reports the base model examining the association between the general fertility

rate and the time preference variables measured by the Sexual Freedom Index (SFI) and the two

ideology variables- Citizen’s ideology and Ranney Index. In all our tables, we begin by reporting

random and fixed effects models. As noted in the previous section, our model can potentially

suffer from simultaneity bias. In column 3, we instrument the independent variables by their first

and second lag values. This approach is done as it is very hard to find external instruments for

these variables. Finally, in column 4, we allow for both endogeneity as well as the dynamic

nature of the determination of fertility by including its lag value as an additional explanatory

variable. This model is estimated using Arrelano-Bond two-step systems estimator. We also

provide various diagnostic tests for IV and dynamic panel data estimators, showing the

appropriateness of the instrument set and model choice. In all our models, we allow for a non-

15
linear relationship between SFI and fertility by including a linear as well as the quadratic term of

the former.

Table 3: Effect of SFI and ideology on General Fertility Rate

(1) (2) (3) (4)
AB
Dependent Var: Gen. Random Fixed Fixed Effects Systems
Fertility Rate Effects Effects IV Estimator±

General Fertility Rate (t-1) 0.287***

(0.091)
Sexual Freedom Index
(SFI) -55.379*** -62.401* -200.772*** 38.182
(19.565) (32.179) (77.546) (71.054)
SFI2 19.908*** 23.788** 72.348*** -12.340
(6.471) (10.197) (24.960) (23.492)
Citizen Ideology -0.191*** -0.012 -0.306 -0.148
(0.050) (0.068) (0.201) (0.108)
Ranney Index 7.490** 14.283*** 15.417** 11.197**
(3.286) (3.637) (7.684) (4.591)

Observations 241 241 239 192

R-squared 0.128 -0.090
Number of stid 48 48 48 48
K-P Underid. test p val 0.0002
Overid. Test p val 0.1803
AR(2) p value 0.256
Hansen Overid. test p val 0.162
Observations 241 241 239 192
Number of stid 48 48 48 48
Robust standard errors in parentheses
±: Windmeijer bias correct robust standard errors reported
*** p<0.01, ** p<0.05, * p<0.1

We start by reporting results from a random effects model. The linear term of SFI is

negative while the square term is positive and significant. This indicates that there is a non-linear

association between SFI and fertility. Increased SFI is associated with a reduction in fertility rate.

Citizen ideology is also negative and significant. Since citizen ideology and SFI are considered

to capture the rate of time preference, this result agrees with our general hypothesis. Ranney

Index has a statistically significant and positive effect on fertility. Random effects models,

16
however, assume that there is no correlation between the state-fixed effects and other state-level

independent variables. The fixed effect model, which relaxes this assumption indicates that SFI

continues to have a negative and non-linear association with fertility, while Ranney Index is

positive and statistically significant. Citizen ideology, however, is no longer statistically

significant. Our main results are in column 3, where we allow for endogeneity of SFI and the

two ideology variables. Citizen Ideology is again not statistically significant. However, SFI

continues to exert a negative effect on fertility, confirming our initial hypothesis. Further, a

higher Ranney Index value suggests that more Democrat control of the state also shows a higher

fertility rate.

In column 4, we estimate a dynamic panel data model. We find that there is positive

feedback of past periods' fertility rates on the current period’s fertility rates. None of the

measures of time preference are significant. Ranney Index is positive and significant.

Table 4: Effect of SFI and ideology on General Fertility Rate, inclusion of state income
(1) (2) (3) (4) (7)
AB
Dependent Var: Gen. Random Fixed Fixed Effects Systems AB Diff
Fertility Rate Efects Effects IV Estimator± Estimator±

General Fertility Rate (t-1) 0.339*** 0.119

(0.116) (0.093)
Sexual Freedom Index
(SFI) -52.453*** -39.903 -120.954** 33.566 16.419
(19.372) (31.989) (54.428) (80.052) (48.759)
SFI2 19.317*** 17.722* 47.726*** -10.592 -2.475
(6.400) (10.077) (18.120) (26.232) (16.410)
Citizen Ideology -0.170*** 0.034 0.084 -0.186** 0.162
(0.050) (0.068) (0.193) (0.087) (0.102)
Ranney Index 3.626 7.875* 1.256 13.057* 8.390
(3.533) (4.003) (11.127) (7.898) (5.486)
Log GSP per capita -2.539*** -3.271*** -4.627** 0.619 -0.954
(0.910) (0.956) (2.194) (2.603) (2.280)
K-P Underid. test p val 0.029
Overid. Test p val 0.0122
AR(2) p value 0.248 0.576
Hansen Overid. test p val 0.4 0.179

17
 Observations 241 241 239 192
Number of stid 48 48 48 48
Robust standard errors in parentheses
±: Windmeijer bias correct robust standard errors reported
*** p<0.01, ** p<0.05, * p<0.1

In Table 4, we include a state-level income measure, a log of gross state product per

capita. In column 1, SFI and Citizen’s Ideology continue to be negative and significant while is

negative and significant. In our main specification reported in column 3, while the two ideology

variables are not significant, SFI continues to have a non-linear and negative impact on fertility

rates. Increasing levels of state economic development are also associated with lowering fertility

rates. In the dynamic panel data specification, however, while the effect of SFI and Income

measures is statistically insignificant, Citizen’s Ideology is again negative and statistically

significant. The magnitude of the lagged fertility variable continues to be rather small. As an

alternative specification, we estimate the dynamic model with Arellano Bond Difference GMM

model. These results are in column 5. The estimate of the lagged fertility term is no longer

significant. This suggests that a static specification of the model given by Equation 1 may be

more appropriate.

In our final specification, we do additional robustness checks by including religion and

race variables in our static panel data estimation. Due to the quality of the data, we only included

the majority groups, shares of Catholics, Protestants, and whites in the state population. Results

are presented in Table 5. SFI continues to have a negative and non-linear effect on fertility, while

Citizen Ideology is only significant in the random effects estimation (column 1). Taken together,

our results seems to confirm Sorokin’s thesis that increasing sexual freedom may result in lower

fertility rates and these results are robust to the inclusion of race and religion variables as well as

after allowing for endogeneity of ideology and SFI variables. In addition, state economic
18
development, measured by per capita income is also associated with lowering fertility rates. The

two ideology variables, as well as race and religion controls, are not statistically significant when

we allow for endogeneity in column 3.

Table 5: Effect of SFI and ideology on General Fertility Rate, additional

controls
(1) (2) (3)
Dependent Var: Gen. Random Fixed Fixed Effects
Fertility Rate Effects Effects IV

Sexual Freedom Index (SFI) -49.945*** -44.599 -117.846**

(18.934) (32.482) (52.579)
SFI2 18.304*** 18.814* 46.756***
(6.296) (10.188) (17.455)
Citizen Ideology -0.131** 0.034 0.134
(0.056) (0.070) (0.208)
Ranney Index 1.366 7.650* 1.214
(3.680) (4.039) (11.579)
Log GSP per capita -3.415*** -3.157 -8.493**
(1.144) (2.374) (3.817)
% of Catholics -12.016* -20.238 -13.863
(6.795) (19.350) (20.118)
% of Protestants -17.567 -12.504 -107.215
(12.278) (52.733) (67.600)
% of whites -0.589 9.905 3.023
(7.042) (16.142) (22.628)
K-P Underid. test p val 0.0266
Overid. Test p val 0.0225
Observations 241 241 239
Number of stid 48 48 48
Robust standard errors in parentheses
±: Windmeijer bias correct robust standard errors reported
*** p<0.01, ** p<0.05, * p<0.1

6. Conclusion
Most studies on fertility use individual-level data. Using aggregate data at the state level

comes with numerous data constraints including the reality that laws about sexual freedom lag

social norms. One persistent result is that citizen ideology negatively impacts the general

fertility rate providing evidence that time preferences play a role in fertility i.e., higher time

19
preference states have less fertility. Our main specification reveals that the SFI is negative and

statistically significant, supporting Sorokin’s thesis that sexual anarchy/freedom leads to lower

fertility. Additionally, our tests for non-linearity, represented by the SFI², show that it is positive

and significant. This indicates that while initial increases in sexual freedom may reduce fertility,

beyond a certain point, further increases might have a different impact. This finding aligns with

studies using individual-level data, suggesting a more complex relationship between sexual

freedom and fertility.

Further empirical research could benefit from improved data, such as a more precise

fertility measure rather than the crude birth rate. The teen fertility rate is another important

variable, but it is only available annually for a limited number of years. Urbanization is another

key variable of interest, but its definition has evolved, and data is only available for specific

years. Similarly, the education variable is only accessible for more recent decades. These two

variables might be proxied by our GSP data. The Ranney Index shows no significance in our

main specification.

20
Referencesa
Arellano, M., & Bond, S. (1991). Some tests of specification for panel data: Monte Carlo evidence and an
application to employment equations. The Review of Economic Studies, 58(2), 277-297.
https://doi.org/10.2307/2297968
Arellano, M., & Bover, O. (1995). Another look at the instrumental variable estimation of error-
components models. Journal of Econometrics, 68(1), 29-51. https://doi.org/10.1016/0304-
4076(94)01642-D
Banfield, E. C. (1974). The unheavenly city revisted. Little, Brown and Company.
Becker, G. S., & Mulligan, C. B. (1997). The endogenous determination of time preference. The Quarterly
Journal of Economics, 112(3), 729-758.
Bellani, D., Arpino, B., & Vignoli, D. (2021). Time preferences and fertility: Evidence from Italy.
Demographic Research, 44, 1185-1228. https://doi.org/10.4054/DemRes.2021.44.50
Belsky, J., Caspi, A., Moffitt, T. E., & Poulton, R. (2020). The origins of you: How childhood shapes later life.
Harvard University Press.
Berry, W. D., Fording, R. C., Ringquist, E. J., Hanson, R. L., & Klarner, C. E. (2010). Measuring citizen and
government ideology in the U.S. states: A re-appraisal. State Politics & Policy Quarterly, 10(2),
117-135. https://doi.org/10.1177/153244001001000201
Berry, W. D., Ringquist, E. J., Fording, R. C., & Hanson, R. L. (1998). Measuring citizen and government
ideology in the American states, 1960-93. American Journal of Political Science, 42(1), 327-348.
https://doi.org/10.2307/2991759
Billari, F. C., & Tabellini, G. (2010). Italians are late: Does it matter? In J. B. Shoven (Ed.), Demography and
the Economy. University of Chicago Press, NBER books.
Blundell, R., & Bond, S. (1998). Initial conditions and moment restrictions in dynamic panel data models.
Journal of Econometrics, 87(1), 115-143. https://doi.org/10.1016/S0304-4076(98)00009-8
Bose, F. (2015). The determinants of sexual freedom from 1990 to 2010. Applied Economics Letters, 1-6.
https://doi.org/10.1080/13504851.2015.1021449
Bose, F. (2021). Policy innovativeness and sexual freedom. Social Science Quarterly, 102(4), 1496-1510.
https://doi.org/10.1111/ssqu.12984
Bose, F., & Jacob, J. A. (2018). Changing sexual regulations in the U.S. from 1990 to 2010: Spatial panel
data analysis. Review of Economics and Institutions, 9(1), 1-18.
https://doi.org/10.5202/rei.v9i1.21418
Bose, F., & Van Duyn, Z. (2020). Time preferences as partisan politics: what do party manifestos show in
twenty-two OECD countries? European Politics and Society, 1-16.
https://doi.org/10.1080/23745118.2020.1847571
Chesson, H. W., Leichliter, J. S., Zimet, G. D., Rosenthal, S. L., Bernstein, D. I., & Fife, K. H. (2006). Discount
rates and risky sexual behaviors among teenagers and young adults. Journal of Risk and
Uncertainty, 32(3), 217-230. https://doi.org/10.1007/s11166-006-9520-1
Clarkson, J. J., Chambers, J. R., Hirt, E. R., Otto, A. S., Kardes, F. R., & Leone, C. (2015). The self-control
consequences of political ideology. Proceedings of the National Academy of Sciences, 112(27),
8250-8253. https://doi.org/10.1073/pnas.1503530112
Craig, L., Powell, A., & Smyth, C. (2014). Towards intensive parenting? Changes in the composition and
determinants of mothers' and fathers' time with children 1992–2006. The British Journal of
Sociology, 65(3), 555-579. https://doi.org/10.1111/1468-4446.12035
Dommermuth, L., Klobas, J., & Lappegård, T. (2015). Realization of fertility intentions by different time
frames. Advances in Life Course Research, 24, 34-46. https://doi.org/10.1016/j.alcr.2015.02.001
Dykstra, P. A., & Fokkema, T. (2011). Relationships between parents and their adult children: a West
European typology of late-life families. Ageing and Society, 31(4), 545-569.
https://doi.org/10.1017/S0144686X10001108

21
Folbre, N. (2008). Valuing children: Rethinking the economics of the family. Harvard University Press.
https://doi.org/10.2307/j.ctvjnrt57
Golsteyn, B. H. H., Grönqvist, H., & Lindahl, L. (2014). Adolescent time preferences predict lifetime
outcomes. The Economic Journal, 124(580), F739-F761. https://doi.org/10.1111/ecoj.12095
Hatemi, P. K., Crabtree, C., & McDermott, R. (2017). The relationship between sexual preferences and
political orientations: Do positions in the bedroom affect positions in the ballot box? Personality
and Individual Differences, 105, 318-325. https://doi.org/10.1016/j.paid.2016.10.008
Hill, M. S. (1988). Marital stability and spouses' shared Time: A multidisciplinary hypothesis. Journal of
Family Issues, 9(4), 427-451. https://doi.org/10.1177/019251388009004001
Holtz-Eakin, D., Newey, W., & Rosen, H. S. (1988). Estimating vector autoregressions with panel data.
Econometrica, 56(6), 1371-1395. https://doi.org/10.2307/1913103
Hoppe, H.-H. (2001). Democracy: The God that failed. Transaction Publishers.
Jones, L. E., Schoonbroodt, A., & Tertilt, M. (2010). Fertility theories: Can they explain the negative fertility-
income relationship? In J. B. Shoven (Ed.), Demography and the Economy. University of Chicago
press, NBER Research.
Klarner, C. (2013). Other Scholars’ Competitiveness Measures https://doi.org/hdl/1902.1/22519
Lewbel, A. (1997). Constructing instruments for regressions with measurement error when no additional
data are available, with an application to patents and R&D. Econometrica, 65(5), 1201-1213.
https://doi.org/10.2307/2171884
Lucas, C. (2018). Fertility: Concepts and measures https://slideplayer.com/slide/12257487/, US Census
and USAID.
Mosley, H. (2006). Fertility and reproduction: Data sources and crude indicators of fertility
http://ocw.jhsph.edu/courses/PopulationChange/PDFs/Lecture4.pdf, Johns Hopkins Bloomberg
School of Public Health.
Mulligan, R. F. (2008). Religion as adaptation: The role of time preference In W. R. Levin (Ed.), Political
Economy Research focus (pp. 69-92). Nova Science Publishers, Inc.
Murphy, M., & Wang, D. (2001). Family-Level continuities in childbearing in low-fertility societies.
European Journal of Population, 17(1), 75-96. https://doi.org/10.1023/A:1010744314362
Murray, M. (2009). Strange bedfellows: Criminal law, family law and the legal construction of intimate life.
Iowa Law Review, 94, 1253-1313.
Nau, M., Dwyer, R. E., & Hodson, R. (2015). Can't afford a baby? Debt and young Americans. Research in
social stratification and mobility, 42, 114-122. https://doi.org/10.1016/j.rssm.2015.05.003
Peters, J., & Büchel, C. (2011). The neural mechanisms of inter-temporal decision-making: understanding
variability. Trends in cognitive sciences, 15(5), 227-239.
https://doi.org/10.1016/j.tics.2011.03.002
Roodman, D. (2009). A note on the theme of too many instruments*. Oxford Bulletin of Economics and
Statistics, 71(1), 135-158. https://doi.org/10.1111/j.1468-0084.2008.00542.x
Skirbekk, V. (2008). Fertility trends by social status. Demographic Research, 18(5), 145-180.
https://doi.org/10.4054/DemRes.2008.18.5
Sorokin, P. A. (1956). The American sex revolution. Porter Sargent Publisher.
Sunde, U., Dohmen, T., Enke, B., Falk, A., Huffman, D., & Meyerheim, G. (2021). Patience and comparative
development. The Review of Economic Studies, 89(5), 2806-2840.
https://doi.org/10.1093/restud/rdab084
Weber, M. (1930 ). The protestant ethic and the spirit of capitalism (T. Parsons, Trans.). Charles Scribner's
Sons. (1905)
Wilson, M., & Daly, M. (2004). Do pretty women inspire men to discount the future? Proceedings of the
Royal Society of London. Series B: Biological Sciences, 271(Suppl 4), S177-S179.
https://doi.org/10.1098/rsbl.2003.0134

22
Windmeijer, F. (2005). A finite sample correction for the variance of linear efficient two-step GMM
estimators. Journal of Econometrics, 126(1), 25-51.
https://doi.org/10.1016/j.jeconom.2004.02.005
Wodarz, D., Stipp, S., Hirshleifer, D., & Komarova, N. L. (2020). Evolutionary dynamics of culturally
transmitted, fertility-reducing traits. Proceedings of the Royal Society B: Biological Sciences,
287(1925), 1-10. https://doi.org/doi:10.1098/rspb.2019.2468
Wright, J. P., Beaver, K. M., Morgan, M. A., & Connolly, E. J. (2017). Political ideology predicts involvement
in crime. Personality and Individual Differences, 106, 236-241.
https://doi.org/10.1016/j.paid.2016.10.062

23
 Biographies

Feler Bose is a professor of economics and finance at Indiana University East in Richmond, Indiana. His
research spans applied microeconomics, political economy, law and economics, and the economics of
religion. He is the author of the book “Sexual Freedom and Its Impact on Economic Growth and
Prosperity.”

Jeffry Jacob is a Professor of Economics at Bethel University in St. Paul, Minnesota. He teaches a variety
of courses in economics, machine learning, and data analytics. Over the past decade, he has conducted
research examining the role of institutions in economic development. His articles have appeared in
journals such as International Review of Economics, Singapore Economic Review, Regulation and
Governance.

24

View publication stats