Economic Growth: Lessons and Puzzles

1
Economic GrowthLessons and Puzzles

• Lant Pritchett
• BYU Economics Faculty
• June 17, 2005

2
Outline Summarize four recent papers then
puzzles and directions

• Education and growth (Does Learning to Add up Add
  up?)
• Labor mobility and growth (Boomtowns and Ghost
  Countries)
• Lessons of the 1990s (with Roberto Zagha and
  team)
• Growth Accelerations (with Rodrik and Hausmmann)

3
Does Learning to Add up Add up?

• Only reason to use aggregate data is to examine
  difference between micro (Mincer) and macro
  returns
• Does education explain much of growth?
• Not really
• Is there evidence of output externalities
• Hugely depends

4
Micro-mincer are pretty consistent, and downward
sloping (r-S)
5
Percentage vs. absolute growth makes a big
difference
6
Bils Klenow on S to SK
7
Variations in assumptions about ? encompass the
log and level versions
8
How about psi?

• If ?0 then (K-L and others)
• But ? is estimated as the slope in the Mincer
  coefficient wrt Sand the t-test of ?0 is over 6

9
Partial scatter plot (conditioning out K/W)my
preferred specification
10
Same regression, assumption ?0
11
Back to the fundamental empirical problem/question
12
Same figure for CUDIE
13
Bottom line Problems with inferring from
aggregate data

• Education expanded at roughly the same pace in
  nearly all countries
• Estimates of TFP are very low, particularly in
  developing countries
• Attributing any significant fraction of output
  growth in fast growing countries to education
  makes TFP negative in others
• Heterogeneity in the impact between rich and poor
  and among poor must be answer

14
Boomtowns and Ghost Countries
15
Variations in growth of population, GDP per capita
16
versus regions within large countries
17
Adjustment in Ireland
18
Adjustment in Bolivia
19
County by county population variation into
contiguous regions
20
Variations within the USA
Region of the United States (contiguous counties) Population 1930 (000s) Percent change in population 1930-1990 Ratio of current population to counter factual at rate of natural increase Number of countries (of 192) with smaller area (with examples) Ratio of area per capita income to national average
Texlahoma 835.8 -36.8 0.31 117/ 192 (Nicaragua, Ban 92.2
Heartland 1482.6 -34.0 0.33 117/ 192 85.2
Deep South 1558.2 -27.9 0.36 96/ 192 (Jordan) 62.6
Pennsylvania Coal 1182.9 -27.9 0.36 43/ 192 (Trinidad and Tobago, Mauritius) 84.5
Great Plains North 1068.0 -27.7 0.36 128/ 192 (Great Britain, Ghana) 85.4
Nation 123202.6 101.9 100.0
21
Five disappointments and one pleasant surprise

• Length, depth, and variance in the transition
  recession in the FSU/EE countries
• Severity and intensity of the financial crises
• Argentinas implosion
• Weakness of the response to reform, particularly
  in Latin America
• Continued stagnation in Sub-Saharan Africa
• Rapid growth in the reforming socialists

22
Transition recession How long? How Deep?

23
Deep, long, and variable
24
Markets miss the mark on East Asia
25
Collapse of convertibility in Argentina

• The Mexican debt crisis marked the end of one
  era, the Argentina crisis perhaps the end of
  another
• The severity given the collapse was not a
  surpriseit was design.
• What was surprising was that it collapsed in
  spite of every attempt to pre-commit

26
Growth came, but did not stay in Latin America,
despite steady reform progress
27
The predicted renaissance in Africa recedes
into the future

• While there are brief episodes of growth in some
  SSA countries, there is no locomotive to pull
  the rest along.
• This is a disappointment, but a surprise?

28
Steady, rapid growth in China, India, and
Vietnam pretty gradualist reformers from
socialism
EAP
SAR
OECD LAC MNA
AFR ECA
29

30
The star performers in growth are middle of the
pack in many indicators
Rank in Control of corruption Regulatory quality Growth
China 63 94 3
Vietnam 105 135 4
India 86 101 14
Out of N 151 180 136
31
Growth Accelerations

• Volatility of growth within countries over time
  is very large
• How common are growth accelerations?
• Defined as period of 7 years
• Acceleration of more than 2
• To more than 2.5
• Exceeds previous peak
• Are there explanations of their timing?

32
Table 1 Episodes of rapid growth classified by growth rates before and after the episode (identified by three letter country acronym and year of acceleration to rapid growth) Table 1 Episodes of rapid growth classified by growth rates before and after the episode (identified by three letter country acronym and year of acceleration to rapid growth) Table 1 Episodes of rapid growth classified by growth rates before and after the episode (identified by three letter country acronym and year of acceleration to rapid growth) Table 1 Episodes of rapid growth classified by growth rates before and after the episode (identified by three letter country acronym and year of acceleration to rapid growth) Table 1 Episodes of rapid growth classified by growth rates before and after the episode (identified by three letter country acronym and year of acceleration to rapid growth)
Growth rate in the ten years from seven years after the initiation of the growth episode (t7 to t17) (with at least 7 years of datano episodes after 1986) Growth rate in the seven years before the initiation of the episode of rapid growth (t, t-7) Growth rate in the seven years before the initiation of the episode of rapid growth (t, t-7) Growth rate in the seven years before the initiation of the episode of rapid growth (t, t-7)
Growth rate in the ten years from seven years after the initiation of the growth episode (t7 to t17) (with at least 7 years of datano episodes after 1986) Negative before (lt0)  (15/69) Slow before (gt0 lt2) (32/69) Above average before (gt2) (22/69)
Growth rate in the ten years from seven years after the initiation of the growth episode (t7 to t17) (with at least 7 years of datano episodes after 1986)    Negative lt0  (after) (16/69) GHA65 GNB69 JOR73 NGA67 TCD73 (slow to growth episode back to slow) ECU70 MLI72 MWI70 RWA75 TTO75  COG78  DZA75  IDN87  PAN75  ROM79  SYR74 (fast to growth episode to slow growth)
Growth rate in the ten years from seven years after the initiation of the growth episode (t7 to t17) (with at least 7 years of datano episodes after 1986)   Slow lt0 gt2 (after) (16/69) DOM69 PAK62 UGA77 ARG63 ZWE64 AUS61 COL67 GBR82 LSO71 NIC60 NZL57 URY74 BRA67  ISR67  PRY74  THA86
Growth rate in the ten years from seven years after the initiation of the growth episode (t7 to t17) (with at least 7 years of datano episodes after 1986)    Above average  gt2 (after) (37/69)         CHL86 CMR72 EGY76 IDN67 MAR58 MUS71 THA57  (slow to growth episode and stays rapid)     CAN62 ESP84 PER59 IND82 PRT85 IRL58 SYR69 IRL85 USA61 KOR62 LKA79 MUS83 CHN78 NGA57 COG69 PAK79 DNK57 PAN59 BEL59 TUN68 BWA69 TWN61 ESP59  FIN58  FIN67  ISR57  JPN58  KOR84  MYS70  SGP69  (fast to growth episode (even faster) to fast)
33
Accelerations

• Large number of episodes of acceleration to rapid
  growth
• The episodes ended very differently (some
  sustained, some crashes)
• Story by story there are some illustrative
  cases (e.g. Indonesia 67, Korea 62, Mauritius 71,
  Chile 88, China 78)
• Regressions have a very hard time predicting
  accelerations with standard reforms

34
Stylized facts of growth

• Steady and near equal growth of leading countries
  over very long-run (2 percent)
• Divergencewith very low measured TFP
  especially in poor countries
• Huge inter-temporal variance in growth within
  countriesmany growth accelerations and
  decelerations
• Possibilitybut rareof sustained miracle
  growth rates

35
Puzzles and directions

• Why such huge heterogeneity in the output impact
  (and hence growth adjustment) from policy reform?
• binding constraints approach
• How to reconcile institutions rule in very long
  (levels) and long (40 years) with stability of
  institutions and growth volatility?
• reform is like a box of chocolates