Diversification: why and how?
description
Transcript of Diversification: why and how?
Diversification: why and how?
Ricardo HausmannHarvard Kennedy School & Center for International Development at Harvard University
A reinterpretation of the sources of growth
Why are some countries rich and others poor?
The traditional explanation
Why are some countries poor and others rich?
The traditional answer emphasizes factor accumulation and “aggregate” productivity
Countries are rich because they work with more: Physical Capital (accumulated through investment) Human Capital (years of schooling)
…and have more “total factor productivity” …which lets them get more output per capita Policy implications:
More education and health More savings to finance investment (and micro-finance) More property rights to assure investors More “productivity”
The neo-classical dead-end
Assume the world is made out of four things Output Physical capital Human capital Labor
85% of growth is explained by non-of-the-above Total Factor Productivity = a measure of our
ignorance (Abramovits, 1958) Maybe it is not that good a way to parse what
the world is made out of
Rich countries do not just produce more per capita
They produce different products
An alternative view
Countries differ in the variety of the capabilities they have
Rich countries make more kinds of products and more complex products that require many capabilities
Poor countries make simple products that reflect the few capabilities they have
The hidden structure in the comparative advantage of nations
The hidden structure in the comparative advantage of nations
It does not matter what level of aggregation you use
Countries Products
Countries Capabilities Products
Intuition
Countries have capabilities Products require capabilities Countries that have more capabilities will be
able to make more products They would be more diversified
Products that require more capabilities will be made by fewer countries Products will be less ubiquitous
Intuition (cont’d)
Countries that have more capabilities will be able to make products that are less ubiquitous
Hence, countries that have more capabilities will be more diversified but will make less ubiquitous products
Diversification of countries and ubiquity of products are negatively correlated They are indirect measures of the capability set of
countries
Method of Reflections: k-k1 diagramAv
erag
e ub
iqui
ty o
f pro
duct
s
Hidalgo CA, Hausmann R Proc. Natl. Acad Sci. (2009) 106(26):10570-10575
Diversification correlates with income per capita
ALB
ARG
ARM
AUS AUT
AZE
BDI
BENBFA
BGD
BHRBHS
BLRBLZ
BOL
BRA
CAF
CAN
CHL
CHN
CIVCMR
COL
CRI
CYPCZE
DEUDNK
DOMDZA
ECU EGY
ESP
EST
ETH
FIN
FJIGAB
GEOGHAGINGMB
GRC
GTMGUY
HKG
HND
HRV
HTI
HUN
IDNIND
IRL
IRN
ISLISR ITA
JAMJOR
JPN
KAZ
KEN
KGZ
KNA
KOR
LBNLKA
LTULVA
MAC
MAR
MDA
MDG
MEX
MLI
MLT
MNG
MOZ
MUS
MWI
MYS
NERNGA
NIC
NLDNOR
NPL
NZL
OMN
PAK
PANPER
PHL
PNG
POL
PRT
PRY
ROMRUS
RWA
SAU
SDNSEN
SGP
SLE
SLV
SVK
SVN
SWE
SYR
TGO
THA
TJK
TKM
TTO
TUR
TZA
UGA
UKR
URY
VENWSM
ZAF
ZMB
ZWE
67
89
1011
0 100 200 300 400k
LYPPPK Fitted values
Diversification
Log
GDP
per c
apita
Hidalgo CA, Hausmann R Proc. Natl. Acad Sci. (2009) 106(26):10570-10575
Hidalgo CA, Hausmann R Proc. Natl. Acad Sci. (2009) 106(26):10570-10575
Complexity in 1985, controlling for initial GDP per capita
What makes growth difficult?
The chicken and egg problem You cannot make new products because you
lack the capabilities You don’t want to accumulate the capabilities
because the products that need them are not being made Because of other missing capabilities
How does the world deal with this? By moving towards “nearby” products
Step 1: Maximum Spanning TreeOur Approach: Distance measured by
probability that, if a country is good in one product, it’s also good in another product.
What is the shape of a forest? Homogenous or
Heterogeneous? What does it look like?
Step 2: Overlay Strong Links
0.4 >
0.4 – 0.55
0.55 – 0.65
0.65 <
CA Hidalgo, B Klinger, A-L Barabasi, R Hausmann.Science (2007)
Using Feenstra et al. Trade Data:
132 Countries1006 Products
SITC-4(1975-2000)
How do monkeys jump?
Malaysia 1975
Malaysia 1980
Malaysia 1985
Malaysia 1990
Malaysia 1995
Malaysia 2000
Key Implications
Monkeys jump short distances So, where you are now in the product space
determines where you can jump In the case of Malaysia, they moved to and
dominated the electronics cluster How are the ECA countries placed in this
space?
Poland (2000): very well connected & diversified (although no presence in electronics)
Russia (2000): Highly peripheral, concentrated in hydrocarbons and raw materials
The Ukraine (2000): somewhere in between- more connected than Russia, not as well connected as Poland
Summarizing this ‘connectedness’ with one number: Open Forest
ALB
ARG
ARM
AUS
AUT
AZE
BDI
BEN
BFA
BGD
BLR
BOL
BRA
CAF
CAN
CHL
CHN
CIV
CMR
COL
CRI
CZEDEUDNK
DOM
DZA
ECU
EGY
ESP
ETH
FIN
GBR
GEO
GHA
GIN
GRC
GTM
HKG
HND
HRV
HTI
HUN
IDNIND
IRL
IRN
ISR
ITA
JAM
JOR
JPN
KAZ
KEN
KGZ
KOR
LBNLKA
LTU
LVAMARMDA
MDG
MEX
MLI MNG
MOZ
MWI
MYS
NERNGA
NIC
NLD
NOR
NPL
NZL
OMN
PAK PANPERPHL
PNG
POL
PRT
PRY
ROM
RUS
RWA
SAUSDN
SEN
SGP
SLESLV
SVK SWE
SYR
TGO
THA
TJK
TKM
TUR
TZA
UGA
UKR
URY
USA
VEN
ZAF
ZMB
ZWE
1112
1314
15ln
open
_for
est1
b
6 7 8 9 10 11lngdppcpppIncome per capita
Ope
n fo
rest
: opt
ion
valu
e of
jum
ping
to o
ther
tree
s
ARG AUS
AUT
BDI
BEL
BEN
BLZ
BOL
BRA CANCHE
CHL
CHN
COL
CPV
CRI
DEU
DMA
DNK
ECU
ESP
FIN
FRAGBR
GMB
GRC
GTM
GUY
HND
IDN
IND
IRL
IRN
ISL
ISR
ITA
JOR
JPN
KNA
KOR
LCA
LUXMAR
MEX
MLT
MUS
MWI
MYS
NER
NIC
NLD
NOR
NZL
PAN
PERPHL
PRT
SAU
SDN
SENSGP
SWE
TGO
THA
TTO
TUN
TUR
TZA UGA
URY
USA
VCT
VEN
ZAF
ZMB
ALB
ARM
AZE
BGR
BLR
CZE
EST
GEO
HRV HUN
KAZ
KGZ
LTULVA
MDAMKD
MNG
POL
ROM
RUS
SVK SVN
UKR
1213
1415
16ln
open
_for
est
6 7 8 9 10 11lngdppcpppIncome per capitaO
pen
fore
st: o
ptio
n va
lue
of ju
mpi
ng to
oth
er tr
ees
Countries in Eastern Europe and Central Asia face very different opportunities
ARM
AZE
BGR
BLR
CZE
EST
GEO
HRV
HUN
KAZ
KGZ
LTU
LVAMDA
MKD
MNG
POL
ROM
RUS
SVKSVN
TJKTKM
UKR
050
0000
1000
000
1500
000
2000
000
open
_for
est1
b
.2 .4 .6 .8 1troughDepth of Transition Recession (% of 1990 GDP per capita reached at trough)
Ope
n fo
rest
in 1
992
Countries with a better-connected export basket did not suffer as deep an output collapse during transition
ARM
AZE
BGR
BLR
CZE
EST
GEO
HRV
HUN
KAZ
KGZ
LTU
LVA MDA
MKD
POL
ROM
RUS
SVKSVN
TJKTKM
UKR
050
0000
1000
000
1500
000
2000
000
open
_for
est1
b
0 1000 2000 3000 4000 5000distance
..and tend to be those closer to the west.
But proximity to the West has no effect on transition beyond its impact on the export mix: open forest trumps distance to Dusseldorf in a cross-country regression
Distance to Dusseldorf
Ope
n fo
rest
in 1
992
Some implications Why do many poor countries not catch up to rich
countries? Because there is no “stairway to heaven” or sequence of nearby
trees that can get them to the denser parts of the product space What causes the “resource curse” (bad performance by
resource rich countries)? Poor connectedness of the resource intensive sectors
Add value to your raw materials? Forward supply linkages vs capabilities Finland
Why do countries fall into protracted slumps? Because their existing export products get into trouble when they are
in a part of the forest where there are no nearby trees Is innovation the solution?
It is about finding profitable excuses to accumulate capabilities that will be used for some other purpose down the road
Space to grow in existing products
Ease
to
jum
p to
new
pr
oduc
ts: o
pen
fore
st
Low High
Low
High
Bridge over troubled waters
Strategic betsLittle space to improve quality and few nearby trees
Stairway to heaven
Parsimonious industrial policyHelp jump short distances to other products
Let it be
It ain’t brokeAmple space to move in all directions
Hey Jude: make it better Competitiveness policyImprove the quality of what already exists
Strategic approaches
Countries in the Policy Plane
ALB
ARGARM
AZEBDI
BEN
BGR
BLZ
BOL BRA
CHL
CHNCOL
CRI
CZE
ECU
EST
GMB
GTM
GUY
HNDHRV
HUNIDNIND
IRN
JORKGZ
KOR
LTULVA
MAR
MDA
MEX
MKD
MLTMNG
MUSMWI
MYS
NER
NIC
PAN
PER
PHL
POLPRTROM
RUS
SAU
SDN
SEN SVKSVN
TGOTHA
TTO
TUN TURTZA
UGA UKRURY
VEN
ZAFZMB
-2-1
01
Eas
e of
Mov
ing
to N
ew P
rodu
cts
-2 -1 0 1 2Room to Upgrade Quality & Grow in Existing Products
Where are the countries?
EBRD Members in the Policy Plane
ALB
ARM
AZE
BGRCZE
ESTHRV
HUN
KGZLTU
LVAMDA
MKD
MNG
POLROM
RUS
SVKSVN
UKR
-2-1
01
Eas
e of
Mov
ing
to N
ew P
rodu
cts
-2 -1 0 1 2Room to Upgrade Quality & Grow in Existing Products
Where are the countries?
ALBARG
ARM
AUS
AUT
AZE
BDI
BEL
BEN
BGR
BLR
BLZ
BOL
BRA CANCHE
CHL
CHN
COL
CPV
CRI
CZE DEU
DMA
DNK
ECU
ESP
EST FIN
FRAGBR
GEO
GMB
GRC
GTM
GUY
HKG
HND
HRVHUNIDN
IND
IRL
IRN
ISL
ISR
ITA
JOR
JPN
KAZ
KGZ
KNA
KOR
LCA
LTU
LUX
LVA
MARMDA
MDV
MEX
MKD
MLT
MNG
MUS
MWI
MYS
NER
NIC
NLD
NOR
NZL
PAK
PAN
PERPHL
POLPRT
ROM
RUS
SAU
SDN
SENSGP
SVK SVN SWE
TGO
THA
TTO
TUN
TUR
TZAUGA
UKR
URY
USA
VCT
VEN
ZAF
ZMB
1213
1415
16O
pen
Fore
st (l
og)
4 6 8 10 12GDP per capita (log)
Countries face different opportunities to jump to other trees
Countries with closer nearby trees grow faster
MOZ
CHN
IND
ETH
NPL
BGD
PAK
NER
MLI
PHL
SDN
KEN
THA
IDN
BDI
HTI
GHALKA
MAR
TGOMDG
BFA
SENMWI
SLE
MNG
UGA
CAF
BEN
DOM
RWA
BGR
TURALB
GIN
KOR
CIV
HNDCOL
GUY
GTM
BRA
URY
JOR
PNGPER
JAM
BLZ
GMB
HUN
MYS
NIC
EGY
CHLKNA
ARG
SYR
MUS
ZAF
PRT
NGA
CRIFJI
IRL
PRY
ESP
CMR
PAN
BOL
MEX
SLVWSM
GBR
ITA
ISR
MLT
JPNAUTSWEFINNLDDNK
HKG
SGP
CYP
GRCUSANZL
ECU
CANVEN
MAC
AUSDZA
NOR
TTO
IRNISL
BHR
GAB
SAU
OMN
-.04
-.02
0.0
2.0
4.0
6e(
G20
YP
CLC
U |
X )
-4 -2 0 2 4e( l20LYPPPK | X )
coef = -.00122659, se = .00153985, t = -.8
OMN
GAB
SAU
IRN
DZA
BHRNGATTO
GMBECU
UGA
ISLGINWSM
RWA
CMR
VEN
BOLSLV
BDI
MWI
FJI
NICSLECAF
PNG
BFA
BENSYR
PRY
EGYGUY
MACBLZ
GHA
TGO
MUS
NOR
CIV
MDG
AUS
CRI
HND
CYP
SENGTM
JAM
PAN
KNACHL
NZL
MLT
JOR
CAN
SDNMYS
MLI
HKG
SGP
MEX
GRC
PERMNG
NER
COL
ZAF
ALB
ARGKEN
DOM
ISRUSA
LKA
FINURYETH
IDN
DNKNLDJPNSWE
BGDPRT
HTI
AUTMAR
IRL
NPL
GBR
BRAITA
ESPTUR
KOR
HUN
PAK
BGRPHL
THAMOZIND
CHN
-.05
0.0
5.1
e( G
20Y
PC
LCU
| X
)
-3 -2 -1 0 1 2e( l20lopen_forest1b | X )
coef = .00801424, se = .0016893, t = 4.74
Open Forest 1985
Aver
age
grow
th 1
985-
2005
Implications
Development is about a stepping stone process to solve the chicken and egg problem
Much of oil and mining is too disconnected to help trigger a transformation process Dedicated rail and ports Few forward and backward linkages
But there are some spillovers The Steam Engine Medellin gold mining and universities
You need not jump from there Adding value to your natural resource endowment is not what Norway,
Australia or Chile do They have other unrelated sectors
Implications
A commitment to diversification is key It implies a commitment to protect the level
and stability of the real exchange rate matters Fiscal stabilization But also a role for monetary policy
But also a willingness of the government to address chicken and egg problems
Especially those related to the provision of public inputs