Japanese inter-regional migrationpatterns affected by 2011 Tohoku

Disaster, analyzed with 2015 JapanPopulation Census

1

Prof. Makoto, Okumura International Research Institute of Disaster Science

IRIDeS, Tohoku University, Sendai, JapanE-mail:mokmr@m.tohoku.ac.jp

Swift restoration gathers peoples attention,adding to the direct impact of the disaster

Hazard and Exposure

Vulnerability

Resilience

Activity Level of Socio-economic

TimeImpact ∝Hazard�Exposure�Vulnerability

Loss

2

We conduct a research to capture the restoration trend through population change based on the Annual migration statistics (Quantitative Analysis)

Presented at IDRiM 2017 in Reykjavik (Aug.25)

Migration flows to be enlarged after a disaster! Hyogo Prefecture with Kobe Earthquake in 1995

3

0"

50000"

100000"

150000"

200000"

250000"

300000"

350000"1979"

1981"

1983"

1985"

1987"

1989"

1991"

1993"

1995"

1997"

1999"

2001"

2003"

2005"

2007"

2009"

2011"

2013"

Hyogo%Pref.�

Birth_Hyg" Death_Hyg"Immig_Hyg" Emig_Hyg"

1996FYImmigration Peak

Emigration Peaks in 94�

95FY

Annual Emigration / ImmigrationNumbers from/to each Prefecture

Statistical Analysis of migration flows4

Enlargements by Disaster(s)Migration free from disaster �Normal Migration Pattern�

Disaster Size (measured by annual R.I.A.: Relative Affected Inhabitants)

regressed

Inhabitant Registration Statistics�1973-2013�

Disaster Statistics by Fire Agency

�

�

Fixed Effect of Panel Data Analysis

Annual R.A.I Number of cases Cumulative %

No Disaster 0 296 14.35%

Small Size D. 0 � ������� 857 55.89%

Medium Size D. 0.0001��� ������ 587 84.34%

Large Size D. 0.001� ����� 276 97.77%

Huge Size D. 0.01� 47 100.00%

5

Categorization of Disaster Size ( )Di,t

�Disaster size is captured by annual relative affected inhabitants.(Number of affected inhabitants / prefecture inhabitants)�Variables are categorized in four different size.

R.A.I is prepared annually for each prefecture (1970-2013)excluding Okinawa before 1974.We estimate the effect up to 3 years before.We did not consider Tohoku 3 Prefectures, because of lack of reliable number of affected people in Fire Agency Statistics.

Estimated Enlargement effects by disaster of different size

6

0.60.70.80.91.01.11.21.31.4

Smal

l sam

e yr

Smal

l 1 y

r b

efo

re

Smal

l 2 y

r b

efo

re

Smal

l 3 y

r b

efo

re

Med

ium

sam

e yr

Med

ium

1 y

r be

fore

Med

ium

2 y

r be

fore

Med

ium

3 y

r be

fore

Larg

e sa

me

yr

Larg

e 1

yr b

efor

e

Larg

e 2

yr b

efor

e

Larg

e 3

yr b

efor

e

Hu

ge s

ame

yr

Hu

ge 1

yr

befo

re

Hu

ge 2

yr

befo

re

Hu

ge 3

yr

befo

re

Emig ImmigR.I.A>0.01

Emigration will shrink after one yearNo negative effect on Immigration• No social decrease of Population

Emigration increase and Immigration decrease Social decrease of population

R.I.A<0.01Effects of

2011 Disaster were not included!

Migration in Fukushima seems different after the Tohoku Earthquake, Tsunami

and Nuclear Accident in 2011

7

Immigration Peak appeared, as well as

Emigration

0"

20000"

40000"

60000"

80000"

100000"

120000"

140000"

160000"

1979"

1981"

1983"

1985"

1987"

1989"

1991"

1993"

1995"

1997"

1999"

2001"

2003"

2005"

2007"

2009"

2011"

2013"

Miyagi&Pref.�

Birth_Myg" Death_Myg"Immig_Myg" Emig_Myg"

0"

20000"

40000"

60000"

80000"

100000"

120000"

1979"

1981"

1983"

1985"

1987"

1989"

1991"

1993"

1995"

1997"

1999"

2001"

2003"

2005"

2007"

2009"

2011"

2013"

Fukushima)Pref.�

Birth_Fksm" Death_Fksm"Immig_Fksm" Emig_Fksm"

Only emigration Peak was appeared

in Fukushima

Let us investigate migration pattern more qualitatively!

TWO NATIONAL STATISTICSMigration Report by Residence Registration

Japan Population Census

Interval Annual 5 years (Questions on migration: once in 10 years)

PublishedContents

Quantitative Info:Numbers of total emigrants and Immigrants of each Prefecture or municipality.

Including Qualitative Info.:Age StructureOrigin and Destination Prefectures

Source Aggregate the residence registration from municipal governments

Exhaustive Survey to the people actually reside. (Direct distribution and collection)

Problem in Fukushima case

Many emigrants do not move registration, in order to keep rights to get support for the people affected by the Nuclear Accidents.

We cannot capture people’s migrations who already return to Fukushima by Oct.2015.

8

NET-MIGRATION ANALYSIS For Age-classes

2

-20000

-15000

-10000

-5000

0

5000

10000

5~ 10~ 15~ 20~ 25~ 30~ 35~ 40~ 45~ 50~ 55~ 60~ 65~ 70~ 75~ 80~ 85~

������

5 year age category�age in period end year�

Fukushima Prefecture 5 years net-migration

(Imigration – emigration)

1995�2000� 2005�2010� 2010�2015�

Negative (loss)

Positive (gain)

Emigration from each Prefecture in the four intervals10

0

200000

400000

600000

800000

1000000

1200000

1400000

1600000

01-A

02H*3

03�"3

04��3

05823

06��3

077�3

08<�3

09'$3

10:K3

11�13

12=3

13&�B

146��3

15#/3

16��3

175�3

187�3

19�)3

20ED3

21�F3

22I�3

23!43

24�C3

25.?3

26�B�

27�G�

28��3

29�;3

30 +�3

31M�3

32�(3

33��3

34��3

35��3

36 �3

37J�3

38!�3

39L43

407�3

41�?3

42E�3

430%3

44��3

45��3

46N��3

47,93

@�>1985-90@�>1995−00@�>2005−10

1985-19901995-20002005-20102010-2015

Fukushima is the only prefecture where number of emigrants grew in between 2010-2015.

Immigration to each Prefecture in the four intervals11

0

200,000

400,000

600,000

800,000

1,000,000

1,200,000

1,400,000

1,600,000

01-A

02H*3

03�"3

04��3

05823

06��3

077�3

08<�3

09'$3

10:K3

11�13

12=3

13&�B

146��3

15#/3

16��3

175�3

187�3

19�)3

20ED3

21�F3

22I�3

23!43

24�C3

25.?3

26�B�

27�G�

28��3

29�;3

30 +�3

31M�3

32�(3

33��3

34��3

35��3

36 �3

37J�3

38!�3

39L43

407�3

41�?3

42E�3

430%3

44��3

45��3

46N��3

47,93

@�>1975-80 @�>1985-90 @�>1995−00@�>2005−10 @�>2010−15

1975-19801985-19901995-20002005-20102010-2015

There are no particular characteristicsin Tohoku Prefectures

Net Immigration to each Prefecture in the four intervals

12

-200000

-150000

-100000

-50000

0

50000

100000

150000

200000

01-B

02I*3

03�"3

04��3

05823

06��3

077�3

08=�3

09'$3

10;L3

11�13

12>3

13&�C

146��3

15#/3

16��3

175�3

187�3

19�)3

20FE3

21�G3

22J�3

23!43

24�D3

25.@3

26�C�

27�H�

28��3

29�<3

30 +�3

31N�3

32�(3

33��3

34��3

35��3

36 �3

37K�3

38!�3

39M43

407�3

41�@3

42F�3

430%3

44��3

45��3

46O��3

47,:3

9A�?1985-909A�?1995−009A�?2005−109A�?2010−15

1985-19901995-20002005-20102010-2015

Fukushima is the only prefecture where net loss of migration grew in between 2010-2015.

Qualitative Analysis focusing on gender and age.NMF: Non-negative Matrix Factorization

• Non negative factorization Method can show several sex-age migration patterns and give weightings for each region.• Each factor may correspond to objective of migration, but we do

now know the number of factors a priori.

13

XE WEF

=

�n:

Reg

ions

x 2

n:R

egio

nsx

2

m�Sex & Age m�Sex & Agek:factors

k:fa

ctor

sXI WI

How different age-groups jointly move with different purpose?

Next, I show the case of 7 factors division.

Factors: sex and age structure14

0.00E+00

1.00E-02

2.00E-02

3.00E-02

4.00E-02

5.00E-02

6.00E-02

7.00E-02

8.00E-02

9.00E-02

M00

M05

M10

M15

M20

M25

M30

M35

M40

M45

M50

M55

M60

M65

M70

M75

M80

M85+

F00

F05

F10

F15

F20

F25

F30

F35

F40

F45

F50

F55

F60

F65

F70

F75

F80

F85+

MiddleWorker1 DependentFamily2 MiddleFamily3Collage4 SingleWorker5 NewJobWorker6SingleWorker7

Male Female

Scores: What’s happen in each prefecture?15

0

500000

1000000

1500000

2000000

2500000

Emig01

Emig03

Emig05

Emig07

Emig09

Emig11

Emig13

Emig15

Emig17

Emig19

Emig21

Emig23

Emig25

Emig27

Emig29

Emig31

Emig33

Emig35

Emig37

Emig39

Emig41

Emig43

Emig45

Emig47

Immig02

Immig04

Immig06

Immig08

Immig10

Immig12

Immig14

Immig16

Immig18

Immig20

Immig22

Immig24

Immig26

Immig28

Immig30

Immig32

Immig34

Immig36

Immig38

Immig40

Immig42

Immig44

Immig46

SingleWorker7 NewJobWorker6 SingleWorker5 Collage4MiddleFamily3 DependentFamily2 MiddleWorker1

Immigration

Emigration

Weightings: What’s happen in each prefecture?16

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Emig01

Emig03

Emig05

Emig07

Emig09

Emig11

Emig13

Emig15

Emig17

Emig19

Emig21

Emig23

Emig25

Emig27

Emig29

Emig31

Emig33

Emig35

Emig37

Emig39

Emig41

Emig43

Emig45

Emig47

Immig02

Immig04

Immig06

Immig08

Immig10

Immig12

Immig14

Immig16

Immig18

Immig20

Immig22

Immig24

Immig26

Immig28

Immig30

Immig32

Immig34

Immig36

Immig38

Immig40

Immig42

Immig44

Immig46

SingleWorker7 NewJobWorker6 SingleWorker5 Collage4 MiddleFamily3 DependentFamily2 MiddleWorker1

Weightings (relative) in Northeastern 10 prefectures17

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

01Ho

kkaido

02Ao

mori

03Iwate

04Miyagi

05Akita

06Yamagata

07Fukushima

08Ibaragi

09Tochigi

10Gu

mma

SingleWorker7 NewJobWorker6SingleWorker5 Collage4MiddleFamily3 DependentFamily2MiddleWorker1

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

01Ho

kkaido

02Ao

mori

03Iwate

04Miyagi

05Akita

06Yamagata

07Fukushima

08Ibaragi

09Tochigi

10Gu

mma

SingleWorker7 NewJobWorker6SingleWorker5 Collage4MiddleFamily3 DependentFamily2MiddleWorker1

Emigration Immigration

Dependent Family 2 moved out from Fukushima

Middle Worker 1, New Job Worker 6 moved in to Fukushima

Strongly observed Factors in Fukushima18

0.00E+00

1.00E-02

2.00E-02

3.00E-02

4.00E-02

5.00E-02

6.00E-02

7.00E-02

8.00E-02

9.00E-02

M00

M05

M10

M15

M20

M25

M30

M35

M40

M45

M50

M55

M60

M65

M70

M75

M80

M85

+

F00

F05

F10

F15

F20

F25

F30

F35

F40

F45

F50

F55

F60

F65

F70

F75

F80

F85+

MiddleWorker1 DependentFamily2 MiddleFamily3Collage4 SingleWorker5 NewJobWorker6SingleWorker7

New Collage Students went out

Dependent Family as mothers and children went out

Young Workers (more male than

Male Female

Fleshly gained results

• We adopt the NMF analysis on a dataset of age-sex matrix prefecture versus immigrants, emigrants and stay in three periods.

• 1995-2000 (including Kobe earthquake?)• 2005-2010• 2010-2015 (including effect of 2011)

19

2010-15 ImmigrationFactor Scores

20

� _�

��9P

W6?

�,?

��?

D>?

�(?

C ?

K�?

30?

FZ?

�=?

L?

2�Q

B�"?

-;?

��?

A"?

C�?

�5?

TS?

�U?

X�?

+@?

�R?

:M?

�Q&

�V&

'?

�J?

�7�?

\�?

4?

��?

% ?

��?

* ?

Y"?

+�?

[@?

C�?

�M?

T!?

<1?

�?

�!?

]� ?

8E?

O� �H N�bNH �Ia�# �.`c�$/d �.`c�[$d )/[^G

Young Family Immigrants are observed anywhere and any time,.Only Exceptional Case is Fukushima 2010-2015.

21

� ���

Factor scores in Fukushima in 3 periods

1995-00�� 2005-10�� 2010-15�� 1995-00�� 2005-10�� 2010-15��

������

� � ��!�� ����� ���"���# ���"���# ����

lDisappear of Young Family’s immigration.lNew Job Immigration increasedlIncrease in “Young Family”

Emigration Immigration

22

� ���

Emigration Immigration

0%

20%

40%

60%

80%

100%

2000 2010 2015 2000 2010 2015 2000 2010 2015

�� �� ������� �������� � �����

�������� ���� �

New Job Workers, Mobility was relatively high in 1995-2000.

KOBE CASE ‘HYOGO Pref.

Emigration Immigration Stay (not move)

Discussion on the result

Unique patterns were observed in Fukushima.• Young Mothers and children decided to emigrate from

Fukushima (Dependent Family 2)• Fear of influence of radioactivity.

• Middle aged (mostly male) workers (Middle Worker1) and young single workers finding the job (New Job Workers6) moved in Fukushima• In order to have works such as disaster recovery or

the radioactive decontamination.• Middle Family and Dependent Family did not enter

Fukushima

• Many households were dismantled.

23

24

FINDINGS

We investigated the effect of disaster on the inter-regional migration in the following 3 years, considering the disaster size.

Small, Meduim, Large Disaster �R.I.A.<0.01�98% cases�Emigration will shrink after one yearNo negative effect on Immigration• No social decrease of Population• External Assistance is not always necessary

Huge Disaster �R.I.A.>0.01�2% of the total cases�Emigration increase and Immigration decrease • Social decrease of population• External Assistance is strongly necessary

25

Future Research Issue

• Consider Disaster type categorization• Consider the external monetary assistance or

designation of “Serious Disaster to be supported”

• Closer investigations, based on the smaller local area data such as municipalities, or different age groups

• Analysis of the effect on economic performance indexes

26