Religions, Fertility, and Growth in South-East Asia David de la Croix - - PowerPoint PPT Presentation

Religions, Fertility, and Growth in South-East Asia

David de la Croix1 and Clara Delavallade2

1IRES, Universit´

e catholique de Louvain

2IFPRI, Washington

October 18, 2016

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Research Question

In most models of the long-run (Malthus, Solow, Lucas), high fertility is detrimental to growth Many religions are supposedly pro-natalist How big is the effect of religion on development through its effect

• n fertility? [measurement question]

How to identify the possible effect on fertility ? How to go from the micro to the macro implications ?

2 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Literature

Microeconometric literature showing some effect of religious affiliation on fertility or education Adsera (Pop. Stud. 2006), Berman et al. (NBER, 2012), Becker and Woessmann (QJE, 2009), Baudin (2014), Chab´ e-Ferret (2014), Lin and Pantano (2015) do not draw quantitative macro consequences Growth/development models with religion Cavalcanti et al. (2007, ET), Strulik (2014), Cervellati et al. (2014) Show how religious norm emerge and affect preferences do not identify the size of effect using microdata do not particularly focus on fertility Growth empirics Cross-country regressions (Barro and McCleary, AJS, 2003) are not robust (Durlauf et. al. JAE, 2012)

3 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Our approach

Full journey from micro estimates to macro simulations a) Auxiliary model. Estimate empirical relationship between fertility and parental background: religion and education from census data. b) Structural model. Micro model of the household. Identify preference parameters to fit the findings of the auxiliary model. c) Counterfactual analysis with growth model. = literature (either micro-demographic estimates, or growth theories, or cross-country regressions)

4 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Our approach

We assume religion impacts preferences → introduces a wedge in the first-order conditions, modifying behavior Alternatively, one can assume religion impacts household technology ex: contraception (Lin and Pantano (2015)) Services to families, including education (Berman et al. 2012) Similar wedges would be introduced in the focs. We cannot really distinguish between the two “explanations”.

5 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Our sample: South-East Asia

6 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Religious Composition

South-East Asia: Common geographical and cultural influences Different religions present in same region of the world Best place to distinguish country fixed effect vs religion fixed effect Main religions in each country:

No Buddh. Hindu Muslim Cath. Prot. Cambodia 96.9 2.1 0.4 Indonesia 1.1 2.4 87.1 2.3 5.8 Malaysia 0.7 24.3 6.7 54.2 2.6 Philippines 0.3 0.1 4.5 83.4 10.6 Vietnam 80.7 10.8 0.0 5.4 0.5 Thailand 0.1 95.4 3.7 0.7

7 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Data

Census data (IPUMS international, various years) Complete fertility Ni of married women aged 45-70, mother’s education Ef

i , father’s education Em i ,

mother’s religious affiliation Rf

i

census fixed effect Ci birth year fixed effect Bi Five levels of education: (i) No school, (ii) Some primary, (iii) Primary cmpl., (iv) Secondary cmpl., (v) University cmpl. → 25 types of couples Ef

i × Em i

Seven religions Rf

i : No religion, Buddhist, Hindu, Muslim,

Catholic, Protestant, Other

8 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Methodology

Pooling different censuses allows to interact the education dummies with the religion dummies → Allows for differential effects of religion depending on the education level

1900 1910 1920 1930 1940 1950 1960 Cambodia Indonesia Malaysia The Philippines Vietnam Thailand

9 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Distribution of Education

Education Men (i) (ii) (iii) (iv) (v) Educ. No Some Primary Secondary University Total Women schooling primary completed completed completed (i) 155,029 89,151 24,542 1,392 113 270,227 (ii) 13,978 109,132 38,078 4,930 541 166,659 (iii) 2,235 16,874 55,567 14,065 2,097 90,838 (iv) 100 1,058 5,234 12,779 3,834 23,005 (v) 17 117 936 3,568 6,581 11,219 Total 171,359 216,332 124,357 36,734 13,166 561,948

Note: unweighted

10 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Auxiliary model

• A. Benchmark:

Ni = βA

1 Ri + βA 2 Ef i × Em i

+ βA

3 Bi + βA 4 Ci + ǫA i

Ef

i × Em i : vector of 25 categorical variables

• B. Effect of religion varies by education level:

Ni = βB

2 Ri × Ef i × Em i

+ βB

3 Bi + βB 4 Ci + ǫB i

Ri × Ef

i × Em i : vector of 7 × 25 = 175 categorical variables

11 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Estimated Fertility by Education groups

Model A - Fertility of Women born 1945 in the Philippines (No relig. + Catholics) Em

i

Ef

i

(i) (ii) (iii) (iv) (v) (i) 4.25 + 0.91 4.78 + 0.91 4.66 + 0.91 4.54 +0.91 4.16 + 0.91 (ii) 4.90 + 0.91 4.82 + 0.91 4.70 + 0.91 4.33 +0.91 3.77 + 0.91 (iii) 4.26 + 0.91 4.65 + 0.91 4.36 + 0.91 4.09 +0.91 3.39 + 0.91 (iv) 4.23 + 0.91 3.89 + 0.91 3.52 + 0.91 3.42 +0.91 3.12 + 0.91 (v) 3.28 + 0.91 2.99 + 0.91 2.75 +0.91 2.83 + 0.91

12 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Effect of Various Religions

fixed effect s.e. Buddhists 0.331 (0.0725) Hindus 0.218 (0.1127) Muslims 0.560 (0.0907) Catholics 0.914 (0.0461) Protestants 1.040 (0.0803) Other religions 0.675 (0.1113) All religions increase fertility significantly (except Hindus) Catholics Protestants > Muslims Buddhists > No religion Hindus

13 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Fertility according to Model B

Fertility of Philip. Women born 1945 – No relig. [Catholic] [Buddhist] [Muslim] (i) (ii) (iii) (iv) (i) 5.58a – 0.43c 5.71a + 0.36c – 0.30 – 0.26b – 0.86a – 0.44b (ii) 4.92a + 0.90a 5.22a + 0.69a + 0.49b – 0.15c + 0.50b – 0.56a (iii) 4.01a + 1.29a 3.65a+ 1.18a + 0.37b + 0.44b + 1.81a + 2.41a (iv) 3.22a + 1.13a 2.88a+ 1.16a + 0.73a + 1.16a + 1.88a + 1.94a

⇒ Gradient fertility–education depends on religions. They seem to prevent fertility from dropping fast when parents’ education rises.

14 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Robustness

– Check that father’s religion is correlated with mother’s religion – Endogeneity of religion ? Use grand-mother religion instead – Impact of religion country dependent? mean effect - jacknife – Poisson or oprobit instead of OLS Density Distribution of Fertility:

2.0e+04 4.0e+04 6.0e+04 8.0e+04 Frequency 10 20 30 Children ever born

15 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Model of the Household

max

st,nt,et,af

t ,am t

ln(ct) + σ ln(dt+1) + γ ln nthη

t+1

s.t. ct = ωhf

t (1 − af t nt) + hm t (1 − am t nt) − st − etnthT

dt+1 = Rt+1st, ht+1 = µt(θ + et)ξ, nt = 1 φ

• af

t am t .

(1) γ: taste for children vs own consumption η: weight of quality ξ: return on education spending θ: exogenous level of public education σ: psychological discount factor ω: female wage φ: time cost parameter 1: male wage

16 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Time allocation

The maximization problem can be decomposed into two steps. First, for some given number of children, parents allocate their time efficiently: min

af

t ,am t

(ωhf

t af t + hm t am t ) nt

subject to (1) This cost minimization problem leads to the following optimal rules (for n < 1/φ): if 1 φ2n2

t

> hm

t

ωhf

t

> φ2n2

t ,

af

t =

• hm

t

ωhf

t

φnt, am

t =

• ωhf

t

hm

t

φnt, if hm

t

ωhf

t

> 1 φ2n2

t

, af

t = 1,

am

t = φ2n2 t ,

if φ2n2

t > hm t

ωhf

t

, af

t = φ2n2 t ,

am

t = 1.

17 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Solution to household problem

If ωhf

t hm t >

θhT 2φηξ 2 then Interior solution: et = 2φηξ

• ωhf

t hm t − θhT

(1 − ηξ)hT , nt = (1 − ηξ)γ(ωhf

t + hm t )

1 + σ + γ 2φ

• ωhf

t hm t + θhT

4φ2ωhf

t hm t − θ2hT2 .

else, Corner solution: et = 0, nt = γ(ωhf

t + hm t )

2(1 + σ + γ)φ

• ωhf

t hm t

.

18 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Possible effects of religion

∃ 2 effects we can identify by looking at fertility Pro-child (γ ↑): leads people to put more weight on children (number & quality) . vs. own consumption. be fruitful and multiply (Gen 1,28):וּבְרוּוּרפּ Pro-birth (η ↓): leads people to put more weight on number children vs. other goods. Abraham, father of a multitude A pro-child religion (high γ) leads to more spending of the two types, while a pro-birth religion (low η) redirects spending from quality towards quantity

19 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Identification

Rise in γ Drop in η

1.0 1.5 2.0 2.5 3.0hf 3.5 4.0 4.5 5.0 5.5 n 1.0 1.5 2.0 2.5 3.0hf 3.5 4.0 4.5 5.0 5.5 n

→ corner regime is key (but well documented, Vogl 2015)

20 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Spending on Quantity and Quality

A pro-child religion (high γ) leads to more spending of the two types, while a pro-birth religion (low η) redirects spending from quality towards quantity: etnthT ωhf

t + hm t

= γηξ 1 + γ + σ for θ = 0 2φ

• ωhf

t hm t n

ωhf

t + hm t

= γ(1 − ηξ) 1 + γ + σ for θ = 0

21 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Structural estimation - Parameters fixed ex ante

1 period = 30 years hT: f, secondary completed Wages by education level: estimation for the Philippines (Luo & Terada) (i) (ii) (iii) (iv) (v) hf 1 1.035 1.07 1.46 2.16 hm 1 1.065 1.13 1.37 1.86 ξ 0.33 ω 0.75 φ 0.065 σ 0.99120 = 0.3

22 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

• Min. distance estimation

Some parameters are fixed a priori min

θ,γz,ηz

• z
• i,j

pi,j,z( ˆ Ni,j,z − n⋆[θ, γz, ηz, hf (i), hm(j)])2. Model B No relig. Catholic Buddhist Muslim θ 0.055

(0.0012)

γz 0.674 0.746 0.621 0.704

(0.0378) (0.0152) (0.0737) (0.0092)

ηz 2.114 1.943 1.872 1.751

(0.0519) (0.0309) (0.0555) (0.0552)

23 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Pro-birth and Pro-child Religions

Non religious Catholics Buddhists Muslims 0.08 0.18 0.28 0.08 0.18 0.28 g (1-h x) /(1+s+g) weight

• n

quantity g h x/(1+s+g): weight on quality Pro-child (D+g) Pro-birth (D- h)

24 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

The corner regime

Parameter θ determines the threshold at which parents shift from no spending in quality of children, to facing a trade-off between quantity and quality if children. Ex: a couple with same human capital, no religion: h > θ 2√ωφηξ hT = 0.055 2 √ 0.75 0.065 × 2.114 × 0.333 hT = 0.69hT. Hence, only couples with human capital at least equal to 69% of the human capital of the teacher will invest in education.

25 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Macro Model

Effect of these differences in preferences on long-run growth ? BGP: ht = hf

t = hm t = hT

Externality (here, endo. growth. For exo. growth, see paper): µt = µ hτ

t hT1−τ.

Production: Yt = AK ε

t L1−ε t

Equilibrium: Lt =

• ωht(1 − φn/√ω) + ht(1 − φn√ω) − etnthT

Pt, Pt+1 = Pt nt/2, Kt+1 = Ptst.

26 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Theoretical results

In the corner regime (no education), A pro-child religion (∆+γ) has a negative effect on income per capita (≈ Solow). A pro-birth religion (∆−η) has no effect beyond making the corner regime more likely. In the interior regime (with endogenous growth), A pro-child religion (∆+γ) has no effect on long-run growth. A pro-birth religion (∆−η) permanently affects the long-run growth rate.

27 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Macro parameters

Parameters τ 1/10 κ 0 or 1 ρ 2% µ g = 1.0230 ε 1/3 A (1 − ε)AK ε

t L−ε t

= 1 Initial conditions: ht/hT = 0.3 Kt such that capital/labor ratio = steady state

28 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Hypothetical countries - endogenous growth

Initial conditions are such that ht/hT = 0.3. Start in the corner regime Only difference across hypothetical countries is η and γ

No relig. affil. Catholics Buddhists Muslims t = 1 nt 5.31 5.67 5.03 5.46 θ + et (% gdp) 4.26% 4.49% 4.08% 4.35% st/((1 + ω)htwt) 15.17% 14.64% 15.59% 15.03% annual growth 3.06% 2.97% 3.15% 3.02% t = 6 nt 3.93 4.40 4.04 4.69 θ + et (% gdp) 8.91% 8.68% 6.42% 5.44% st/((1 + ω)htwt) 15.17% 14.64% 15.59% 15.03% annual growth 2.24% 1.85% 1.71% 1.48%

29 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Uncertainty surrounding the estimates

10 20 30 40 50 Non religious Catholics Buddhists Muslims Non religious Catholics Buddhists Muslims Endogenous growth Exogenous growth

GDP per cap. in the Hypothetical Economies after 6 Periods: Confidence Intervals

30 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Countries: 1950-80

countries’ growth rates growth gaps Vie- Tha- Ind- Tha- Ind Mal Phi Vie Tha

• Ind
• Phi
• Phi
• Ind

data 1950-80 2.85 2.88 2.69 0.47 3.87

• 2.38

1.18 0.15 1.02 endogenous growth t=1 3.02 3.06 2.97 3.07 3.14 0.04 0.17 0.05 0.12

Matches relative performance countries (but Vietnam). 10% to 30% of magnitude is explained by religion.

31 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Countries: 1990-2010

countries’ growth rates growth gaps Vie- Tha- Ind- Tha- Ind Mal Phi Vie Tha

• Ind
• Phi
• Phi
• Ind

data 1980-2010 3.09 3.44 0.81 4.94 4.43 1.85 3.62 2.28 1.34 endogenous growth t=2 2.16 2.19 2.29 2.51 2.29 0.34 0.00

• 0.12

0.12 t=3 1.78 1.81 2.00 2.25 1.93 0.47

• 0.07
• 0.22

0.15

Religion explains – lead of Vietnam ( 1/5 of the difference) – domination of Thailand over Muslims countries (10% of the gap) Failure: cannot explain the bad performance of the Philippines

32 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Savings

st = σ 1 + σ + γ (ωhf

t + hm t ),

γ assumed religion specific, σ assumed constant across religions A) Do differences in γ as identified from the fertility behavior influence the savings behavior? House ownership as a Function of Theoretical Savings st: Linear probability model Coef.

• Std. Err.

t P > |t| st .6526 .1765 3.70 0.014

Note: N = 510994, R2 = 0.85, Std. Err. clustered by country, Census & year fixed effects included

33 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Savings (2)

B) was it reasonable to assume that σ (thrift) does not depend on religious affiliation? Linear probability model dependent variable: residual of house ownership regression Coef.

• Std. Err.

t P > |t| buddhist

• .0768

.0160

• 4.81

0.005 muslim

• .0390

.0280

• 1.39

0.223 catholic

• .0296

.0208

• 1.42

0.214

Note: N = 510994, R2 = 0.02, Std. Err. clustered by country, Census & year fixed effects included, Reference group includes individuals with no religious affiliation

→ Buddhists are saving less than expected

34 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Quality of education

Country

• Math. PISA 2012

Science PISA 2012 et at t = 4 Vietnam 511 528 5.8 Thailand 427 444 5.1 Malaysia 421 420 4.8 Indonesia 375 382 4.8 Cambodia NA NA 5.1 Philippines NA NA 6.4

35 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Conclusion

Pro-natalist religions can or cannot damage growth, depending on – the stage of growth – whether they are pro-child (∆+e, n) or pro-birth (∆−e, ∆+n) One can identify these effects by looking at how religion and education interact in explaining fertility From South-East Asian censuses, Islam is the most pro-birth while Catholicism is the most pro-child Account for 10% of the gap between buddhists and muslims countries over 1980-2010 With secularization, one may think that these effects will be weaker in the future

36 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Hypothetical countries - exogenous growth

No relig. Catholics Buddhists Muslims t = 1 (et = 0) nt 4.62 5.07 5.56 5.1 yt 0.95 0.89 0.98 0.92 annual growth 2.03% 1.91% 2.09% 1.95% t = 6 (et > 0) nt 3.16 4.01 3.77 4.49 yt 29.21 23.74 25.52 21.56 annual growth 2.22% 2.13% 2.11% 2.07%

36 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Robustness to other measures of return to schooling

Country year male female prim. sec. tert. prim. sec. tert. Cambodia 2004 5 3.1 14 11.8 4 16.6 Indonesia 2010 9.6 8.7 12.6 12.7 12 12.9 Malaysia 2010 7.6 9.3 21.8 6.8 12.3 23.1 Philippines 2011 7 6.4 20.1 3.7 6.1 29.4 Thailand 2011 2.7 4.6 16.6 1.4 5.9 19.2 Vietnam NA

(i) (ii) (iii) (iv) (v) hf 1 1.21 1.48 2.23 3.76 hm 1 1.26 1.60 2.70 4.82

36 / 36

Introduction Auxiliary Structural Counterfactuals Further implications Conclusion Supplements

Pro-birth and Pro-child Religions with High Returns to Schooling

Non religious Catholics Buddhists Muslims 0.08 0.18 0.28 0.08 0.18 0.28 g (1-h x) /(1+s+g) weight

• n

quantity g h x/(1+s+g): weight on quality Pro-child (D+g) Pro-birth (D- h)

36 / 36