Download Socialist Growth Revisited: Insights from Yugoslavia

Socialist Growth Revisited:
Insights from Yugoslavia
Leonard Kukić
London School of Economics, Department of
Economic History
Introduction (I)
• Successor states of Yugoslavia have essentially stagnated over the
past 30 years.
• Diverging European economic development after WWII had
attracted much public interest. Not academic interest necessarily.
• Planned economies performed relatively well in the 1950s and the
1960s.
• In the 1980s their performance turned dismal.
• (Usual) Explanation: Growth based on capital and labour expansion,
and the transfer of resources from farms to factories, intrinsically
limited (Krugman, 1995).
Introduction (II)
• Failure of planned economies mostly attributed to
embedded inefficiencies.
• Doom argument: Employers and employees faced poor
incentives since property was state owned (Bardhan
and Roemer, 1993).
• Nuanced argument: performance was relatively OK
during mass production technology of the
1950s/1960s. But bad with onset of flexible production
technologies during late 1970s (Broadberry and Klein,
2011) .
Motivation (I)
• Existing literature suffers from two problems.
• 1st Problem: Excessive focus on Soviet Union.
• Masks heterogeneity of E. European countries.
• I analyse Yugoslavia.
• Yugoslavia was taken as an example of one of fastest growing countries in
1950-70s.
• Thus, Balassa and Bertrand (1970) found Yugoslavia did better than the
average of other 9 sample countries, in terms of output and TFP growth.
• In AER, Horvat (1971) attributed this to Yugoslavia’s decentralised
economic system.
Motivation (II): 1980s
Motivation (III): The evolution of macroeconomic indicators in Yugoslavia and in the U.S., 1952-90.
Figure 1.2: Capital to labour ratio
90
70
50
1990 Int. GK$, log scale
1990 Int. GK$, log scale
Figure 1.1: GDP per capita
70
50
30
10
5
2
1952
1960
1970
Year
1980
30
10
5
2
1952
1989
0.7
0.4
0.6
0.3
0.2
0.1
0
1952
1970
Year
1980
1989
Figure 1.4: Hours worked per capita
0.5
Hours per person
Ratio
Figure 1.3: Investment to output ratio
1960
0.5
0.4
0.3
1960
1970
Year
1980
1989
0.2
1952
Yugoslavia
1960
1970
Year
1980
U.S.
Note: Capita is defined as working age person, while labour is defined as total hours
worked ((average yearly hours worked per employee) x (total number of employees)).
Hours worked per capita are total hours worked divided by the working age population.
1989
Motivation (IV)
• 2nd Literature problem: Typically relies on a simple
comparison of macro indicators, like productivity and
growth (van Ark, 1996; Broadberry and Klein, 2011).
• Hence, arguments not quantified or tested.
• Some growth accounting exercises (Balassa and
Bertrand, 1970; Vonyo, 2010). Very useful, but growth
acc. suffers from its own problems.
• So what can be done?
Motivation (V)
•
I apply business cycle accounting (BCA). Developed by Cole and Ohanian (2002)
and Chari et al. (2007), among others.
•
BCA is a diagnostic tool like growth accounting, but moves towards explanations.
•
Hence, I can “explain” both the success and failure episodes.
• BCA composed of two steps:
1. Calculate from data.
2. Insert wedges into a prototype model to determine their impact on econ.
growth.
• As a dynamic general equilibrium (DGE) model, confers two major advantages.
1. Adds a timing dimension.
2. Identifies the incentives that drive output, capital and labour.
Preview of results
• 1. TFP became more important over time in sustaining growth.
Reconciles conflicting results in literature about relative importance
of factors and TFP (Balassa and Bertrand, 1970; Weitzman, 1970;
Sapir, 1980; Bergson, 1983; Kontorovich, 1986).
• 2. The labour wedge consistently deteriorated since the mid-1960s.
And drove the collapse of growth during the 1980s.
• Similar to findings of Weitzman (1970), Sapir (1980), and Easterly
and Fischer (1995). But on completely different grounds.
• Does not mean that technology and diminishing returns on capital
were un-important. Rather, labour frictions were more important.
History
• Evolution of socialist economic system can be divided into four phases.
Gradual move from central planning, through market socialism, to
decentralised planning.
• 1st Phase, 1947-1951: Rigid central planning focused on heavy
industrialisation. Soviet Union Template.
• 2nd Phase, until 1965; Yugoslav officials sought to distance themselves
from Soviet Union. Yugoslavia began gradually opening towards the West.
• 3rd Phase, until 1975: 1965 reform important. Heyday of market socialism.
Economic power further decentralised to work councils within firms.
• 4rd Phase, until the end: 1974 constitution led to further decentralisation
of power to the level of departments within firms.
History IV: Trade as per cent of GDP (1990 Int. GK$), and composition of trade, 19521988.
90
80
70
50
40
30
20
10
0
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
%
60
Year
Trade as % of GDP
OECD as % of total trade
Socialist Bloc as % of total trade
Note: This measure of openness can be considered real, since GDP is PPP adjusted. Trade means exports and
imports of goods and services. Composition of trade, however, reefers to composition of trade in goods, due to
data constraints.
History V: Western aid as per cent of GDP and gross investments (1990 Int. GK$),
1952-1965.
12.00
10.00
%
8.00
6.00
4.00
2.00
0.00
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
Year
% of gross investments
% of GDP
1962
1963
1964
1965
Methodology
• BCA based on a standard Ramsay-Cass-Koopmans
growth model.
• Chari et al. (2007) argue that a large set of DGE models
can be simplified through addition of “wedges”.
• BCA developed for accounting for business cycle
fluctuations. But can be applied to study episodes of
economic growth (Lahiri and Yi, 2009; Lu, 2012;
Chakraborty and Otsu, 2013; Cheremukhin et al., 2014.
Intuition (I)
• BCA cannot identify the policies that effect the economy, bur rather the
evolution of incentives that firms and households face.
• Four wedges used are channels through which policies affect growth.
• Taken together, they drive economic growth, and match data.
• Labour wedge is related to incentives that determine the supply of labour.
• Increasing labour wedge can be interpreted as an increase in the return
for work effort.
• Capital wedge is related to incentives that determine savings and
investments, both physical and human capital.
• Increasing capital wedge can be interpreted as an increase return on
capital.
Intuition (II)
• Income wedge embodies aggregate demand
shocks stemming from G and NX.
• Efficiency wedge (TFP) measures the efficiency
with which inputs are transformed into output.
• Drawbacks:
1. Wedges do not interact.
2. Can’t identify the exact incentives.
Prototype model: Setup
Utility f.:
Budget constraint:
Production function:
Capital law of motion:
Equilibrium with wedges (I)
• Efficiency wedge: .
As in RBC models, is measured as the deviation around
(labour augmenting tech. progress).
• Labour wedge: .
• Measures discrepancy between and MPL.
Equilibrium with wedges (II)
• Capital wedge:
(1+𝛾)𝑐𝑡+1
𝑐𝑡 𝛽
+𝛿−1=
𝑦𝑡+1
𝜃
𝑘𝑡+1
1 − 𝜏𝑘,𝑡+1 .
Measures the discrepancy between intertemporal
substitution and the real interest rate.
• Income wedge: 𝑦𝑡 − 𝑐𝑡 − 𝑖𝑡 = 𝑦𝑡 𝜏𝑖,𝑡 .
Measures the expenditure gap for the resource
constraint to hold.
Data (I)
•
Official output series problematic (Social Product).
•
Services excluded (education, healthcare, government, and etc.).
•
But input from excluded services into other sectors included.
•
Gross inconsistency in the application of the Material Planning System.
• Furthermore, official output growth inflated due to:
1. Index number problems (Gerschenkron, 1947).
2. Distorted prices (Staller, 1986).
3. Perhaps outright fabrication.
•
As such, I use output series from Maddison (2010). But created by Thad Alton et al.
(1970, 1992).
Data (II)
• Working age population (15-64), employment of social sector, and
data on private farming employment taken from official sources.
• Total yearly labour input de-trended by 3600.
• Human capital initially approximated by average years of schooling
from Barro and Lee (2013). I take their estimate for Serbia.
• Avg. years of schooling turned into mincerian human capital as in
Hall and Jones (1999).
• Capital stock series problematic. Exclude an investment category
called “other”.
Calibration
• Assume 𝛽 is 0.95, and assume that 𝜙 is 2, as in similar
countries (Lu, 2012; Cheremukhin et al., 2014).
• 𝜃 is 0.4, as in Easterly and Fischer (1995). Kukic (2015)
finds 0.43 for Yugoslavia.
•
•
•
•
Remaining parameters taken from data.
𝜐𝑡 is time-varying, and on average 1.1 % per annum.
𝛾 is 0.9 % (constant).
𝛿 is 5.46 %, to ensure modelled capital stock matches
the 1990 data.
Assumptions
• 1990 is the terminal period of wedges.
• Profit maximisation is a poor description of socialist firms.
But a socialist economy can be seen as a heavily distorted
version of a perfectly competitive economy.
• Cobb-Douglas assumption of unit substitution between
capital and labour is problematic (Weitzman, 1970; Sapir,
1980; Easterly and Fischer, 1995).
• Might be below one. If so, provides an elegant explanation
for socialist growth - planned economies ran into acute
diminishing returns on capital.
Results: The evolution and interpretation of wedges
Figure 4.2: Capital wedges
Frictionless benchmark model =1
Steady State in benchmark =100
Figure 4.1: TFP
280
260
240
220
200
180
160
140
120
100
1952
1960
1970
Year
1980
1989
1.4
1.2
1
0.8
0.6
0.4
1952
1960
1970
Year
1980
1
0.5
0
1952
1960
1970
Year
1980
1989
Figure 4.4: Income wedges (as a share of GDP)
0.5
1.6
Percentage of output (Y)
Frictionless benchmark model =1
Figure 4.3: Labor wedges
1.5
1989
Yugoslavia
0.4
0.3
0.2
0.1
1952
1960
1970
Year
1980
1989
U.S.
Note: Business cycles have been cycled out using the Hodrick-Prescott filter (smoothing
parameter = 6.25). No technological growth rate is imposed (γ = 0), rendering TFP
growth comparable to standard growth accounting exercises.
Interpreting TFP (I)
• Nishimizu and Page (1982) argue that TFP was
driven by efficiency rather than technology.
Similar to Hsieh and Klenow (2009).
• Viable interpretations:
• 1. Reconstruction dynamics (Vonyo, 2008).
• 2. Structural change or improvements in
sectoral allocation of resources (Lewis, 1954;
Vollrath, 2009).
Year
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
1979
1978
1977
1976
1975
1974
1973
1972
1971
1970
1969
1968
1967
1966
1965
1964
1963
1962
1961
1960
1959
1958
1957
1956
1955
1954
1953
1952
Share
Interpreting TFP (II): Share of agricultural workers in
total workforce in Yugoslavia, 1952-89
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
Interpreting TFP (III)
• Trade might had boosted output through TFP
(Alcala and Ciccone, 2004).
• Yugoslavia did specialise according to its
comparative advantage.
Interpreting labour wedge (I)
Hall (1997) argues labour wedge reflects frictions that lead households to spend a
long time on non-market activities.
Figure 6.1: Total hours worked
18
In billions
17
16
15
14
1952
1960
1970
Year
1980
1989
1980
1989
Figure 6.2: Working age population
17
16
15
In millions
•
14
13
12
11
10
1952
1960
1970
Year
Interpreting labour wedge (II)
• Mismatch between total hours worked and
the working age population reflected in
increasing unemployment.
• Unemployment rate an average 8.2 per cent
during 1967-75, rose to an average 12.6
percent during 1976-87.
• Migration patterns not helpful.
Interpreting labour wedge (III)
• Chari et al. (2007) argue that labour wedge can
reflect distortions caused by monetary
contraction (deflation) and trade unions (nominal
wage rigidity).
• In Yugoslavia during 1980s, real money balances
halved (Bradley and Smith, 1991).
• Labour managed firms under-invested, to pay out
high(er) wages (Estrin, 1983).
Interpreting labour wedge (IV): Unit wage cost in
efficiency units
150
140
130
110
100
90
80
70
1987
1986
1985
1984
1983
1982
1981
1980
1979
1978
1977
1976
1975
1974
1973
1972
1971
1970
1969
1968
1967
1966
1965
1964
1963
1962
1961
1960
1959
1958
1957
1956
1955
1954
1953
60
1952
Index, 1952 = 100
120
Year
Note: Unit wage cost is the ratio of the wage per worker to the GDP per efficiency unit of labour (labour productivity augmented by
technology). For each year, the said ratio is divided by the same ratio of 1952. Wage rate has been deflated using the official output
deflator.
Interpreting labour wedge (V): Labour unrest in
Yugoslavia, 1958-89
1958
1978
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
Frequency
of strikes*
Number of
strikes
Number of
workers on strike
2.8
30
62
47
18
96
100
104
163
227
228
232
n/a
n/a
235
216
174
336
393
696
851
1685
n/a
n/a
n/a
n/a
13,504
13,507
10,997
21,776
29,031
60,062
88,860
288,686
n/a
n/a
Media
reports of
strikes
n/a
n/a
3
8
24
36
86
158
195
734
320
n/a
Note: *1980 to 1989 shows data for Slovenia, a member republic of Yugoslavia.
Source: Stanojević (2003) for the frequency of strikes; Jovanov (1989) for the number of strikes and the
number of workers involved; Lowinger (2009) for media reporting of strikes
Results (I): The contribution of wedges to economic growth
1952-1960
Average annual Growth rate
TFP
Capital wedge
Labour wedge
Income wedge
1960-1970
Average annual Growth rate
TFP
Capital wedge
Labour wedge
Income wedge
1970-1980
Average annual Growth rate
TFP
Capital wedge
Labour wedge
Income wedge
1980-1989
Average annual Growth rate
TFP
Capital wedge
Labour wedge
Income wedge
Output per working
age person
Labour
Capital to labour
ratio
5.6
5.0 (90%)
2.0 (36%)
0.3 (5%)
1.32 (24%)
-1.3
0.5 (38%)
-1.6 (-123%)
-6.6 (-508%)
-2.1 (-162%)
4.6
1.6 (35%)
5.6 (122%)
5.2 (113%)
4.2 (91%)
3.3
2.6 (79%)
1.2 (36%)
0.2 (6%)
1.5 (45%)
-7.1
-3.9 (-55%)
2.0 (28%)
-6.2 (-87%)
1.0 (14%)
8.3
4.1 (49%)
3.1 (37%)
3.8 (46%)
8.2 (99%)
3.7
4.7 (127%)
0.0 (0%)
0.0 (0%)
3.2 (86%)
-3.6
0.0 (0%)
-4.4 (-122%)
-4.4 (-122%)
6.3 (175%)
6.3
5.5 (87%)
1.3 (21%)
1.8 (29%)
2.4 (38%)
-1.2
0.5 (38%)
1.0 (83%)
-0.9 (-75%)
0.0 (0%)
-2.5
-2.2 (-88%)
2.4 (96%)
-3.9 (-156%)
-0.8 (-32%)
1.0
4.6 (460%)
-0.9 (-90%)
0.4 (40%)
-0.7 (-70%)
Results (II): The actual evolution of GDP per capita versus the
counterfactual evolution of it (without TFP), 1952-89
Yugoslavia: GDP per working age person
400
Index, 1952=100, log scale
300
200
100
1952
1960
1970
Year
Without TFP
1980
1989
Actual
Notes: The 1952 level of GDP per working age person is indexed to 100. If the
two lines move in parallel, it means that the combined capital, labour and
income wedges are responsible for most of economic growth.
Results (III): Simulations of GDP per working age person versus
the actual GDP per working age person, 1952-89
Yugoslavia: GDP per working age person
400
Index, 1952=100, log scale
300
200
100
1952
1960
Capital wedge only
1970
Year
TFP & capital wedge
1980
TFP, capital & labor wedges
1989
Actual
Conclusion (I)
• I hope I had filled a knowledge void.
• TFP became more important.
Reconciles conflicting finding in the literature.
• Labour frictions were the most important drag on growth.
Reconfirms older findings.
• Natural step forward:
Determine the quantitative causality between policies and
TFP and labour frictions.
Conclusion (II)
• For TFP, trade might have been important.
• More research needed to understand the incentive/ability
of households to provide work effort.
Largely ignored so far.
• Wages, driven by behaviour of labour-managed firms,
might had led to deterioration of labour wedge in the
1960s and the 1970s, but not the 1980s.
• Unemployment, and potentially labour unrest, seem
important for the 1980s.
Appendix slides
Baseline wedges: Simulations of GDP per
working age person versus the actual GDP per
working age person, 1952-89
Yugoslavia: GDP per working age person
400
Index, 1952=100, log scale
300
200
100
1952
TFP only
1960
Capital wedge only
1970
Year
Labor wedge only
1980
Income wedge only
1989
Actual
Baseline wedges: Actual and simulated
I/Y, 1952-89
Non-agriculture: The actual evolution of Nonagricultural GDP per capita versus the counterfactual
evolution of it (without TFP), 1952-89
Yugoslavia: GDP per capita
500
400
Index, 1952=100, log scale
300
200
100
1952
1960
1970
Year
Without TFP
1980
Actual
1989
Data parameters (Beta = 0.93, phi = 4.02): Simulations
of GDP per working age person versus the actual GDP
per working age person, 1952-89
Yugoslavia: GDP per working age person
600
500
Index, 1952=100, log scale
400
300
200
100
1952
1960
Capital wedge only
1970
Year
TFP & capital wedge
1980
TFP, capital & labor wedges
1989
Actual
Data parameters (Beta = 0.93, phi = 4.02): The actual evolution
of GDP per capita versus the counterfactual evolution of it
(without TFP), 1952-89
Yugoslavia: GDP per working age person
600
500
400
Index, 1952=100, log scale
300
200
100
1952
1960
1970
Year
Without TFP
1980
Actual
1989
Linear leisure utility function: Simulations of GDP per working age
person versus the actual GDP per working age person, 1952-89
Yugoslavia: GDP per working age person
400
Index, 1952=100, log scale
300
200
100
1952
1960
Capital wedge only
1970
year
TFP & capital wedge
1980
TFP, capital & labor wedges
1989
Actual
Linear leisure utility function: The actual evolution of
GDP per capita versus the counterfactual evolution of it
(without TFP), 1952-89
Yugoslavia: GDP per working age person
400
Index, 1952=100, log scale
300
200
100
1952
1960
1970
Year
Without TFP
1980
Actual
1989
Stone-Geary utility function: Simulations of GDP per working age
person versus the actual GDP per working age person, 1952-89
Yugoslavia: GDP per working age person
400
Index, 1952=100, log scale
300
200
100
1952
1960
Capital wedge only
1970
Year
TFP & capital wedge
1980
TFP, capital & labor wedges
1989
Actual
Stone-Geary utility function : The actual evolution of GDP per
capita versus the counterfactual evolution of it (without TFP),
1952-89
Yugoslavia: GDP per working age person
400
Index, 1952=100, log scale
300
200
100
1952
1960
1970
Year
Without TFP
1980
Actual
1989