Agricultural Trade Policy in Africa: Taxation, Protection

Download Report

Transcript Agricultural Trade Policy in Africa: Taxation, Protection

Explaining Food
and Agricultural Policy:
New Data and Hypothesis Tests
Will Masters
[email protected]
www.agecon.purdue.edu/staff/masters
Friedman School Seminar
24 September 2008
Motivation: the development paradox
Motivation: the development paradox
New Data
• A 3-year project funded through the World Bank involving
100+ researchers and case studies for 68 countries, 77
commodities over 40+ years
• Project results to be published in six books
– Four volumes of country narratives
• Africa (Anderson & Masters); Asia (Anderson & Martin); LAC (Anderson &
Valdes); European Transition (Anderson & Swinnen)
– Two global volumes
• One with regional syntheses and reform simulations
• One with political economy explanations for policy choices
– Results today are mostly from W.A. Masters and A. Garcia (2009), “Agricultural
Price Distortion and Stabilization: Stylized Facts and Hypothesis Tests,” in K.
Anderson, ed., Political Economy of Distortions to Agricultural Incentives.
Washington, DC: World Bank.
• All available at www.worldbank.org/agdistortions
Country coverage
No. of
countries
Africa
16
Percentage of world
Pop.
10
GDP
1
Ag.GDP
6
Asia
12
51
11
37
LAC
8
7
5
8
ECA
13
6
3
6
HIC
19
14
75
33
Total
68
91
95
90
Commodity coverage
(top 30 products only)
No. of
Products
Percentage of world
Production
Exports
Cereal Grains
10
84
90
Oilseeds
6
79
85
Tropical crops
7
75
71
Livestock products
7
70
88
Total
30
75
85
The method: price distortions from
“stroke of the pen” policies
• Tariff-equivalent Nominal Rate of Assistance
in domestic prices relative to free trade:
NRA 
• Sometimes estimated directly from observed policy:
Pd  Pf
Pf
NRA  t
• More often imputed by price comparison:
Pd  (1  m)  E*  Pf
• We also introduce a new “stabilization index”,
for the standard deviations
around trend prices:
sd ( Pˆ f )  sd ( Pˆd )
SI 
sd ( Pˆ f )
100
Explaining the data
Our approach is to test for:
(1) stylized facts
– persistent correlations with broadly-defined variables, that
could result from many different policymaking mechanisms
(2) specific political-economy mechanisms
– other correlations with narrowly-defined variables, as
implied by particular theories of policymaking
– these could explain residuals and add explanatory power to
the stylized facts, or explain the stylized facts themselves
– most tests are weak; only in one case do we have a strong
identification strategy
The three stylized facts
The three broad influences we capture are:
(1) A development paradox from taxation to subsidies as
incomes rise, as measured by real GDP per capita at PPP
prices (PWT 6.2)
(2) An anti-trade bias from taxation of both imports and
exports, as measured by whether commodity is importable
or exportable in each year
(3) A resource curse effect from taxation of natural resources,
as measured by arable land area per capita (FAOSTAT)
Seven specific hypotheses
We test for each standard theory of policy failure:
– Rational ignorance when per-person effects are small
– Free ridership when groups of people are large
(versus more political support from larger groups)
– Rent-seeking by unconstrained incumbents (versus
checks-and-balances from institutions and markets)
– Revenue motives for cash-strapped governments
– Time consistency of policy when taxation is reversible but
investment is not (as opposed to simultaneous choices)
– Status-quo bias from loss aversion or conservative social
welfare functions in politics
– Rent dissipation from the entry of new farmers (as
opposed to free riding among existing farmers)
Results:
A new view of the development paradox
National average NRAs by real income per capita, with 95% confidence bands
Tradables
Net taxation of consumers
NRA>0
0.0
-1.0
-0.5
NRA
0.5
1.0
1.5
All Primary Products
≈$5,000/yr
Export taxes with
import restrictions =
anti-trade bias
NRA<0
Net taxation of farmers
6
(≈$400/yr)
8
10
6
Income per capita (log)
All Primary Products
Exportables
8
(≈$3,000/yr)
10
(≈$22,000/yr)
Our tests aim
to account for
nonlinearity in
these lines,
and also
dispersion
around them,
as well as the
NRA-income
relationship
itself
Importables
Notes: Each line shows data from 66 countries in each year from 1961 to 2005 (n=2520), smoothed with confidence
intervals using Stata’s lpolyci at bandwidth 1 and degree 4. Income per capita is expressed in US$ at 2000 PPP prices.
Results:
A new view of policy change over time
Average NRAs for all products by year, with 95% confidence bands
ASIA (excl. Japan)
ECA
-1
0
1
2
AFRICA
1960
1980
1990
2000
LAC
2
HIC
1970
-1
0
1
Increased taxes on
consumers in 1990s
1960
1970
1980
1990
Heavy taxes on
consumers in the
1980s, then reform
2000
1960
1970
1980
1990
2000
Heavy taxes on
farmers in 1970s
then reform
All Primary Products (incl. Nontradables)
Results:
A new view of policy change over time
Average NRAs for importables and exportables by year, with 95% confidence bands
ASIA (excl. Japan)
ECA
-1
0
1
2
AFRICA
1960
LAC
1980
1990
Trend away from
taxes on exports,
with rising
import restrictions
-1
0
1
2
HIC
1970
1960
1970
1980
1990
2000
1960
1970
Importables
1980
1990
Heavy taxes on
exports in 1970s
2000
then reform
Exportables
with varied
import restrictions
2000
Results:
The stylized facts in OLS regressions
Table 1. Stylized facts of observed NRAs in agriculture
Explanatory variables
Income (log)
Land per capita
Africa
Asia
Latin Am. & Car. (LAC)
High inc. cos. (HIC)
Importable
Exportable
Constant
R2
No. of obs.
(1)
(2)
0.3420***
0.3750***
-0.4144***
-2.6759***
0.28
2,520
-2.8159***
0.363
2,269
Model
(3)
0.2643***
-0.4362***
0.0651
0.1404***
-0.1635***
0.4311***
-2.0352***
0.418
2,269
(4)
(5)
0.2614***
0.2739***
-1.9874***
0.827
2,520
0.1650*
-0.2756***
-2.0042***
0.152
28,118
Notes: Covered total NRA is the dependent variable for models 1-4, and NRA by commodity for model 5. Model
4 uses country fixed effects. Results are OLS estimates, with significance levels shown at the 99% (***),
95% (**), and 90% (*) levels from robust standard errors (models 1-4) and country clustered standard errors
(model 5). The omitted region is Europe and Central Asia.
Source for all tables and charts: W.A. Masters and A. Garcia (2009), “Agricultural Price Distortion and
Stabilization: Stylized Facts and Hypothesis Tests,” in K. Anderson, ed., Political Economy of Distortions to
Agricultural Incentives. Washington, DC: World Bank.
Results:
Specific hypotheses at the country level
Table 2. Hypothesis tests at the country level
(1)
(2)
Total NRA for: All Prods. All Prods.
Explanatory variables
Income (log)
0.2643*** 0.1234***
Land per capita
-0.4362*** -0.2850***
Africa
0.0651
0.1544***
Asia
0.1404*** 0.2087***
LAC
-0.1635*** -0.0277
HIC
0.4311*** 0.2789***
Policy transfer cost per rural person
-0.0773*
Policy transfer cost per urban person
-1.2328***
Rural population
Urban population
Checks and balances
Monetary depth (M2/GDP)
Entry of new farmers
Constant
-2.0352*** -0.9046**
R2
0.4180
0.45
No. of obs.
2,269
1,326
(3)
(4)
(5)
(6)
(7)
All Prods. |All Prods.| Exportables Importables All Prods.
0.3175***
-0.4366***
0.0964**
0.1355***
-0.1189***
0.4203***
0.1913***
-0.4263***
0.2612***
0.1007**
-0.0947***
0.3761***
0.2216***
-0.7148***
-0.1071***
-0.1791***
-0.2309***
1.0694***
0.1142***
-0.6360***
-0.0628
0.0217
-0.1780***
0.8807***
0.2461***
-0.4291***
0.0844**
0.1684***
-0.1460***
0.4346***
1.4668***
-3.8016***
-0.0173***
-0.0310*** -0.0401***
-2.4506*** -1.2465*** -1.5957***
0.437
0.294
0.373
2,269
1,631
1,629
-0.4652*
0.397
1,644
-0.0737*
-1.8575***
0.419
2,269
Notes: Dependent variables are the total NRA for all covered products in columns 1, 2, 3 and 7; the absolute value of that NRA in column
4, and the total NRA for exportables and importables in columns 5 and 6, respectively. For column 2, the sample is restricted to countries
and years with a positive total NRA. Monetary depth is expressed in ten-thousandths of one percent. Results are OLS estimates, with
robust standard errors and significance levels shown at the 99% (***), 95% (**), and 90% (*) levels.
Results:
Specific hypotheses at the product level
Table 3. Hypothesis tests at the product level
Explanatory variables
(1)
Income (log)
0.2605**
Importable
0.0549
Exportable
-0.2921***
Land per capita
-0.3066***
Africa
0.0553
Asia
0.2828
LAC
-0.0652
HIC
0.2605*
Perennials
Time consistency
Animal Products
Others
Lagged Change in Border Prices
Lagged Change in Crop Area
Constant
-1.8516*
R2
0.1950
No. of obs.
25,599
(2)
0.2989***
0.0048
-0.3028***
-0.3352***
-0.1315**
0.2589***
-0.1764**
Model
(3)
0.2363**
-0.0061
-0.2918***
-0.3478***
0.1171
0.2998
-0.0309
0.3388**
-0.1492***
0.2580***
-0.1956**
Status-quo bias
-2.0109***
0.2100
20,063
-1.6685*
0.2240
20,063
(5)
0.3160**
0.1106
-0.3614***
-0.4738***
0.0554
0.1833
-0.1426
0.4837*
(6)
0.2804**
0.0331
-0.3414***
-0.1746**
0.1236
0.2311
-0.0863
-0.0298
-0.0025***
-2.1625**
0.3020
15,982
0.0083
-2.0549*
0.1940
9,932
Notes: The dependent variable is the commodity level NRA. Observations with a lagged change in border prices lower than -1000% were
dropped from the sample. Results are OLS estimates, with clustered standard errors and significance levels shown at the 99% (***), 95%
(**), and 90% (*) levels.
Results:
How much stabilization is achieved?
Stabilization index over the 1961-2005 period, by income level
Exportables
When
stabilizing,
SI>0
0
100
All Primary Products
SI
-200
-100
SI<0
if gov’t is
destabilizing
Non Tradables
-200
-100
0
100
Importables
6
8
10
6
Income per capita (log)
Africa
Rest of the World
8
10
Results:
Richer countries stabilize more
Table 4. Determinants of the stabilization index
Model
Explanatory variables
Income (log)
Importable
Exportable
Land per capita
Income growth variation
Exchange rate variation
Africa
Asia
Latin America
High income countries
Constant
R2
No. of obs.
Dropped obs.
(1)
5.6507***
(2)
6.5568*
1.5545
-37.7412***
0.029
757
20
4.6606**
0.005
766
11
(3)
7.0059***
-7.1127
-8.4469**
-9.8402**
-40.9054**
0.035
722
6
(4)
7.4730***
-9.4289*
-9.5703**
-9.4037**
-444.8959
2.0297***
-44.9126**
0.047
722
6
(5)
9.4113***
8.2332
15.2604**
-4.4882
-3.0503
-75.4189***
0.032
771
6
(6)
8.8422*
-10.3265*
-11.6999**
-9.6186**
-547.3185
1.0391
1.1559
6.2383
-10.931
-1.5757
-53.9286
0.055
724
4
Notes: Dependent variable for all regressions is the Stabilization Index by country and product. Influential outliers were dropped
from the sample based on the Cook's distance criteria [(K-1)/N]. Results are OLS estimates, with clustered standard errors and
significance levels shown at the 99% (***), 95% (**), and 90% (*) levels.
More results:
Since 1995, policies have
moved closer to free-trade prices
National average NRAs by income level, before and after the Uruguay Round agreement
Exportables
Importables
1
0
-1
NRA
2
3
All
Flatter curves, closer to zero
6
7
8
9
10
6
7
8
9
10
6
Income per capita (log)
1960-1994
1995-2005
7
8
9
10
Low-income Africa taxes farmers less,
Higher-income Asia taxes consumers less
National average NRAs by income level, before and after the Uruguay Round agreement
AFRICA, Exportables
AFRICA, Importables
ASIA, Exportables
ASIA, Importables
Pro-farm reforms in
lower-income Africa
NRA
-1
0
1
2
3
AFRICA, All
Pro-consumer
reform in higherincome Asia
-1
0
1
2
3
ASIA, All
6
7
8
9
10
6
7
8
9
10
6
Income per capita (log)
1960-1994
1995-2005
7
8
9
10
There has been less improvement in
E. Europe-Central Asia or Latin America
National average NRAs by income level, before and after the Uruguay Round agreement
ECA, Exportables
ECA, Importables
LAC, All
LAC, Exportables
LAC, Importables
-1
0
1
2
3
NRA
-1
0
1
2
3
ECA, All
6
7
8
9
10
6
7
8
9
10
6
Income per capita (log)
1960-1994
1995-2005
7
8
9
10
AFRICA, Exportables
AFRICA, Importables
The biggest change has been
in high-income countries
-1
0
1
2
3
AFRICA, All
National average NRAs by income level, before and after the Uruguay Round agreement
HIC, All
HIC, Exportables
HIC, Importables
1
-1
0
NRA
2
3
US, EU and Japan: reforms and WTO commitments
6
7
8
9
400
1,000
3,000
8,000
10
22,000
6
7
8
9
400
1,000
3,000
8,000
10
22,000
6
7
8
9
400
1,000
3,000
8,000
10
22,000
Income per capita (log)
1960-1994
1995-2005
But current events could change the pattern:
…will return of high food prices cause policy reversals?
…how will the 2008 credit crisis affect policy choices?
Some conclusions
• Three stylized facts help explain policy choices:
– A development paradox from taxing farmers to taxing
consumers as incomes rise
– An anti-trade bias from taxation of both imports and exports
– A resource abundance effect against natural resources
• Three mechanisms help explain the income effect:
– Rational ignorance when per-person costs are small
– Improved governance from more checks and balances
– Revenue motives for import taxes when financial systems
are deeper
More conclusions
• Four other mechanisms help add to the income effect:
–
–
–
–
More people in the sector leads to more favorable policies
An end to entry of new farmers leads to more farm support
Crops with more sunk costs (perennials) are taxed more
Policy changes try to reverse the last year’s price changes
• Two widely-held views are not supported:
– Policy changes do not try to reverse changes in area planted
– Policy provides little price stabilization in poor countries
 Status quo bias and price stabilization are not
consistent characteristics of real-life policies;
other explanations work better.
Finally…
• Policy relationships have changed over time
– Relative to income levels, prices are now much closer to
free trade than in the past, especially in Africa, Asia and
the high income countries.
• The recent move to freer trade could be reversed
– In particular, a return of 1970s-style food prices could
easily cause a return to 1980s-style food policies.
• Policy outcomes are far from predetermined!
– Our models explain less than half of the variation we see.