PreprintPDF Available

Optimal Agricultural Policy: Small Gains?

Authors:
Preprints and early-stage research may not have been peer reviewed yet.

Abstract and Figures

Agricultural subsidies distort the allocation of workers across sectors, and may keep too many workers in agriculture. We use a general equilibrium model with endogenous sector selection calibrated to the U.S. economy to assess the efficiency loss and redistribution effect of the current transfer system. Eliminating current subsidies has two main effects: (1) small efficiency gains (around 4% of agricultural output), and (2) a corresponding rise in the price of agricultural goods. We find high-productivity farmers to be the main beneficiaries of the existing policies, although some of the transfers generate a redistribution effect towards low-productivity agents, which extends beyond the agricultural sector.
Content may be subject to copyright.
Optimal Agricultural Policy: Small Gains?
Kai DingFilippo Rebessi
April 9, 2020
Abstract
Agricultural subsidies distort the allocation of workers across sectors, and may keep
too many workers in agriculture. We use a general equilibrium model with endogenous
sector selection calibrated to the U.S. economy to assess the efficiency loss and redis-
tribution effect of the current transfer system. Eliminating current subsidies has two
main effects: (1) small efficiency gains (around 4% of agricultural output), and (2) a
corresponding rise in the price of agricultural goods. We find high-productivity farm-
ers to be the main beneficiaries of the existing policies, altough some of the transfers
generate a redistribution effect towards low-productivity agents, which extends beyond
the agricultural sector.
JEL Classification: H21, H25, H30, J24, J31, J43
Keywords: Optimal Transfers, Agricultural Subsidies, Agricultural Wage Gap
We thank three anonymous referees, Wesley Blundell, V. V. Chari, Jed DeVaro, Larry Jones, Joseph
Kuehn, Ryan Lampe, Christopher Phelan, Christian Rossler, Jung You, and the seminar participants at the
CJP Workshop at the University of Minnesota and the Department of Economics seminar at CSUEB for
valuable discussions and comments. Any remaining errors are our own. Authors’ declarations of interest:
none.
California State University, East Bay (e-mail: kai.ding@csueastbay.edu)
California State University, East Bay (e-mail: filippo.rebessi@csueastbay.edu)
1
1 Introduction
The agricultural sector is the recipient of large policy transfers in most developed nations. In
OECD economies, estimated support to farmers averaged 20% of gross agricultural revenues
in 2017, with countries like Iceland, Norway, South Korea, and Switzerland well above the
50% mark. In the United States, support to agricultural producers was 10% in 2017, with
an aggregate amount roughly equivalent to the expenditure on unemployment insurance.
These sector-specific transfers may distort the allocation of workers between agriculture
and the rest of the economy. In this paper, we analyze transfer schemes that eliminate these
distortions in a general equilibrium model, and assess their welfare effect for society.
We find that eliminating distortions would lead to modest efficiency gains. The main
reason for this finding is a general equilibrium effect. Eliminating agricultural transfers
results in a direct loss of income for agricultural workers, which induces them to reallocate
out of the agricultural sector, and reduces the supply of the agricultural good. As a result,
the relative price of the agricultural good increases, compensating agricultural workers for
the initial loss. According to our findings, the distribution of workers across agriculture and
the rest of the economy in the U.S. is fairly close to the efficient one, even with the current
system of agricultural transfers in place.1
We also analyze the redistribution effect of current agricultural policy. Highly productive
farmers, who receive the vast majority of the government payments, are the main benefi-
ciaries. In addition, the transfers exert a moderating effect on the agricultural price, which
tends to benefit lower-income agents the most.
We build on a general equilibrium version of the Roy model of selection with two sectors
and Stone-Geary preferences introduced by Lagakos and Waugh (2013).2In the data, agri-
1It is important to point out that our analysis is not taking into consideration spillover-effects to indus-
tries connected to the agricultural sector, such as agricultural equipment and machinery, food-processing,
and transportation. The extent to which these linkages are significant may imply larger inefficiencies and
misallocation due to the current transfers system.
2Given that the focus of our analysis is a developed economy, we follow Herrendorf and Schoellman (2018)
and Lagakos and Waugh (2013), and take the view that the allocation of workers across sectors is the outcome
of selection. Other recent studies that use micro-level evidence (see Alvarez (2018) and Hicks et al. (2017))
2
cultural transfers come under a variety of categories.3We broadly classify them into two
main groups in our model: transfers proportional to current level of production (proportional
subsidies, henceforth) and transfers disconnected from it (lump-sum subsidies, henceforth).
Furthermore, to be consistent with the empirical observation that the majority of agricul-
tural transfers are received by large farms (Cai (2019)), the rate of proportional subsidy is
allowed to take on two different levels, depending on agricultural productivity. This allows us
to capture the regressive nature of the current system while staying relatively parsimonious
in model parameters.
We calibrate our model to the U.S. economy, and conduct four sets of counterfactual
policy experiments to identify the welfare and redistribution effects of the two forms of
subsidies.
First, we remove proportional and lump-sum subsidies one at a time. Both reforms cause
a fraction of agricultural workers to reallocate outside the sector and a rise in the agricul-
tural price. When proportional subsidies are removed, there is little impact on agricultural
employment, the income gap between agriculture and non-agriculture, as well as welfare, for
most of the farmers. On the other hand, removing lump-sum subsidies leads to a decline in
agricultural employment and a shrinking income gap. In addition, we find that lump-sum
subsidies have a significant redistribution effect towards lower-income agents, which extends
beyond the agricultural sector. This is due to the relative inferiority of the agricultural good,
which implies that lower-income agents are spending a disproportionately higher portion of
their income on it.
Second, we measure the efficiency loss from the current subsidies by replacing them with
non-distortionary counterparts. We design the non-distortionary transfers to be welfare-
find no significant barriers to labor mobility across sectors. There is a large body of literature investigating
the causes of the measured income gap between the agricultural sector and the rest of the economy across
countries. Several studies emphasize the role of barriers to reallocate outside of agriculture (for example,
barriers to adopt modern intermediate inputs and access to the labor market, as in Restuccia et al. (2008)).
3We use data on Producer Support Estimate, which is an indicator of the annual monetary value of gross
transfers from consumers and taxpayers to support agricultural producers. Further details are provided in
Section 3.
3
neutral for all agents, allowing us to identify the gain in production efficiency in isolation
from the redistribution effect of the current policy. We find that both proportional and
lump-sum subsidies are causing an efficiency loss that amounts to around 2% of agricultural
GDP. The reason for the relatively small loss in efficiency lies behind a general equilibrium
effect which works through the adjustment in the agricultural price.
Finally, we analyze two budget-neutral alternative policies as a replacement of the current
system: a consumption subsidy on the agricultural good, and a uniform lump-sum transfer4.
When current subsidies proportional to income are replaced with consumption subsidies, the
agricultural price rises, while the after-subsidy price falls. This indicates a benefit to both
agricultural workers and consumers, with the exception of farmers at the top of the income
distribution. The uniform lump-sum transfer achieves a similar outcome in terms of welfare,
although it hurts high-productivity farmers considerably more.
1.1 Related Literature
U.S. agricultural policy has evolved over time from one based on supply controls and high
price supports to one mostly based on direct government payments (see Gardner (2008) and
Dimitri et al. (2005)). Agricultural and resource policies have been previously analyzed in
computable general equilibrium models (see Hertel (2002) for a review of this approach).
Hertel and Tsigas (1988) studies the effects of eliminating farm and food tax differential
treatments in 1977, and finds that these policies lowered food costs by $4.5 billion in the
same year. Kilkenny and Robinson (1990) analyzes unilateral and multilateral agricultural
liberalization in the United States with a 10-sector computable general equilibrium model,
and finds that “economy-wide gains from agricultural liberalization arise primarily from the
movement of labor out of agriculture”.5
Our policy analysis builds on the model introduced by Lagakos and Waugh (2013), which
4We are indebted to two anonymous referees for suggesting these counterfactual exercises.
5The incidence of agricultural subsidies on land markets and values has also been extensively investigated
in the literature. See Kirwan (2009), Ciaian et al. (2010) and Goodwin et al. (2011).
4
is part of a large literature investigating the large productivity differences in agriculture
across countries (see for example Caselli (2005), Gollin et al. (2014), Cai and Pandey (2015),
Restuccia et al. (2008), and Chen (2017)). Their model combines Roy (1951)’s model of
selection with the use of non-homothetic preferences, which are common in the literature on
structural change.6We add in this model government transfers to the agricultural sector,
for which there is readily available data for a cross-section of developed countries from the
OECD database on agricultural support (see OECD (2018)). The interaction of selection
with policy and institutional frictions is also analyzed in Adamopoulos et al. (2017): the
paper uses farm-level panel data from China in a quantitative framework and shows that
the misallocation effect of distortionary policies can be greatly amplified by selection. Cai
(2019) also analyzes the role of policy distortions on farm size in the United States, and finds
that the increase of regressive distortions occurred since 1970 can account for the increasing
dominance of large farms.
In other related work, Teignier (2018) studies the impact of eliminating trade barriers on
structural transformation in a small open economy with two sectors, and finds that openness
to trade significantly accelerates the transition out-of-agriculture, which is in line with the
large changes produced by our model when the relative price of agriculture is fixed.
Our paper relates also to the extensive literature in political economy studying inefficient
forms of redistribution, of which agricultural support policy is usually the textbook example.
Several papers point to the role of the government’s lack of commitment to rationalize
such policies. Staiger and Tabellini (1987), Dixit and Londregan (1995), and Acemoglu
(2001) all develop models in which the time-consistent government policy has the effect of
“locking” workers into a relatively unproductive sector. Our results suggest that transfers
to agriculture can also be rationalized in a different fashion, that is a form of redistribution
to low productivity workers in any sector in the economy.
The remainder of this paper is organized as follows. We set up our model and define a
6See for example Caselli and Coleman II (2001). Matsuyama (2008) provides a survey of this literature,
while Buera and Kaboski (2012) analyzes the shortcomings of traditional explanations of structural change.
5
competitive equilibrium in Section 2. Section 3documents the magnitude and time-series
pattern of the current transfers to the agricultural sector, and illustrates the mapping of the
model to the data. Section 4analyzes the economic effects and the welfare impact of an
optimal policy that eliminates distortions, as well as other counterfactual policy experiments.
Section 5concludes.
2 Model
We adopt the two-sector framework introduced by Lagakos and Waugh (2013). In this model,
workers select into either agriculture or the rest of the economy based on their comparative
advantage, which is given by their productivities in each sector. We augment the model by
introducing agricultural transfers, which distort the labor allocation.
2.1 Agents
The economy is populated by a continuum of infinitely-lived agents of unit measure, indexed
by i. Each agent supplies one unit of labor inelastically, and is endowed with a pair of
productivity levels (za
i, zn
i).za
irepresents the efficiency units of labor an agent can provide
if they work in the agricultural sector a, and zn
iis the efficiency units of labor if they work
in the rest of the economy n. In each period t, agents choose which sector to allocate their
labor to maximize income yj
it,j∈ {a, n}.(za
i, zn
i)are drawn at birth from a distribution G,
which is modeled, following Lagakos and Waugh (2013), as a Frank-Coupla of two Frechet
distributions characterized by variance parameters θaand θn. The joint distribution of the
individual productivities reads
G(za, zn) = C[F(za), F (zn)]
where
F(zj) = ezjθj
6
and
C(u, v) = 1
ρln (1 + (eρu 1)(eρv 1)
(eρ1) ).
Agents are hand-to-mouth, and enjoy consumption of agricultural and non-agricultural goods
denoted by caand cn, with Stone-Geary preferences
ln(ca
i¯a) + νln(cn
i).
¯ais a minimum consumption requirement for the agricultural good, which makes it relatively
inferior: the share of total consumption allocated to ca
idecreases as agent’s income grows.7
Using the non-agricultural good as the numeraire, pa
tis used to denote the relative price of
the agricultural good. The budget constraint in each period is
pa
tca
it +cn
it max {ya
it, yn
it}.
2.2 Firms
2.2.1 Non-Agricultural Sector
We assume that the non-agricultural sector operates a constant returns to scale production
technology converting labor into non-agricultural output. The profit maximization problem
of a non-agricultural firm is given by
max
Ln
t
AtLn
twn
tLn
t
subject to
Ln
t=Ziωn
t
zn
idG (za
i, zn
i),
where ωn
tis the set of workers selecting into sector nat time t, and Atis an exogenous process
representing the overall productivity level of the economy. Competitive labor markets imply
7This non-homothetic preference is a common assumption of the literature on structural change. See
Matsuyama (2008) for example.
7
that each efficiency labor unit receives a non-agricultural wage of wn
t=At.
2.2.2 Agricultural Sector
The agricultural sector uses a constant returns to scale production technology converting la-
bor into the agricultural good. Agents selecting into agriculture (farmers henceforth) receive
transfers from the government, which we model as a lump-sum payment Tt, in addition to
a component proportional to their output τt. The distinction is motivated by the data on
agricultural support: certain transfers depend directly on the current level of production (for
example, price-support payments and input subsidies), while others are disconnected from
it (for example, land conservation programs).8Furthermore, in order to capture the fact
that larger farmers receive most of the government’s transfers (see Cai (2019)), we allow the
proportional component τtto be dependent on the farmer’s productivity za
i. In particular,
τtcan take two values: (τtL, τtH ). All farmers with productivity below a certain threshold ¯
za
t
receive a proportional subsidy τtL to their labor income, while those above receive τtH . The
profit maximization problem in the agricultural sector at time treads
max
La
t
Atpa
tLa
twa
tLa
t
subject to
La
t=Ziωa
t
za
idG (za
i, zn
i),
where ωa
tis the set of agents selecting into sector aat time t.
We assume that the agricultural transfer system (τt(za
i), T )is financed with a proportional
tax on income ttpaid by all the agents in the economy. Agent i’s income in agriculture and
the rest of the economy are given by
ya
it = (1 tt){Atpa
t(1 + τt(za
i)) za
i+Tt}
8Details on how we map the model to the data are provided in Section 3.
8
yn
it =At(1 tt)zn
i.
The sector-selection problem in each period reads
max
Iit∈{0,1}Iit {Atpa
t(1 + τt(za
i)) za
i+Tt}+ (1 Iit)Atzn
i.
2.3 Competitive Equilibrium
Definition 1. A competitive equilibrium for this economy consists of a price system {pa
t, wa
t, wn
t}t,
allocations {ca
it, cn
it}i,t , sector selection choices {Iit ∈ {0,1}}i,t, and government policies
{tt, τ (za
i), T }tsuch that:
Agents maximize utility subject to the budget constraint
max log(ca
it ¯a) + νlog(cn
it)
s.t. pa
tca
it +cn
it max {ya
it, yn
it}
ya
it = (1 tt){wa
tza
i(1 + τt(za
i)) + Tt}
yn
it = (1 tt)wn
tzn
i
Workers choose the sector with higher after-subsidy labor income
Iit =
1if wa
tza
i(1 + τt(za
i))+Ttwn
tzn
i
0otherwise
Firms maximize profits
wa
t=Atpa
t
wn
t=At
9
Goods markets clear
Zca
itdG (za
i, zn
i) = AtZIit=1 za
idG (za
i, zn
i)
Zcn
itdG (za
i, zn
i) = AtZIit=0 zn
idG (za
i, zn
i)
The Government budget is balanced
ttZmax {wa
tza
i(1 + τt(za
i)) + Tt, wn
tzn
i}dG (za
i, zn
i)
=τtL ZIit=1,za
i<¯
za
t
wa
tza
idG (za
i, zn
i) + τtH ZIit=1,za
i¯
za
t
wa
tza
idG (za
i, zn
i) + TtZIit=1 dG (za
i, zn
i)
3 Data and Calibration
In this section, we describe different forms of agricultural transfers, and how they are mapped
to our model. Data are retrieved from the “Agricultural Policy Monitoring and Evaluation”,
an annual report prepared by the OECD. This report offers up-to-date estimates of support
to agriculture, defined as “the annual monetary value of gross transfers to agriculture from
consumers and taxpayers, arising from governments’ policies that support agriculture, re-
gardless of their objectives and their economic impacts”. One of the main indicators in the
report is the Producer Support Estimate (PSE henceforth). PSE represents policy transfers
to agricultural producers, measured at the farm gate and expressed as a share of gross farm
receipts. Data on PSE, which we use as our measure of agricultural transfers, are available
from 1986 to 2017. Figure 1demonstrates how PSE has evolved over time. The left panel
measures PSE as a fraction of total farm receipts in OECD countries (solid blue line) and
the United States (dashed red line). A quick glance at the time series shows that agricultural
support has been substantial across developed economies. Although decreasing over the past
30 years, PSE still averages around 20% of total farm receipts in OECD countries in 2017,
and around 10% in the United States.
10
Figure 1 – Magnitude of Producer Support Estimate
1990 1995 2000 2005 2010 2015
Year
5
10
15
20
25
30
35
40
Percentage
% PSE in USA and OECD
OECD
USA
1990 1995 2000 2005 2010 2015
Year
0
20
40
60
80
100
120
140
Billions of $
PSE and Unemployment Insurance
PSE
Unemployment Insurance
To offer an idea of the overall magnitude of agricultural transfers, the right panel of
Figure 1compares aggregate PSE with the annual spending on unemployment insurance
in the United States. PSE was roughly 40 billion dollars in 2017, which is similar to the
expenditure on unemployment insurance.
3.1 Measuring Agricultural Transfers
PSE includes a wide variety of budgetary transfers. We broadly classify them into two main
categories: those proportional to current level of production (τt(za
i)in the model) and those
disconnected from it (Tt).
In particular, the OECD splits PSE into five groups of policies: (1) Output Based,(2) In-
put Based,(3) Area/Animal/Receipts/Income (AARI), current,(4) Area/Animal/Receipts/Income
(AARI), non-current, and (5) Payments based on non-commodity criteria. Policies (1), (2),
and (3) are payments proportional to the current level of production. As a result, we include
them in τt(za
i).9(4) and (5) contain transfers that are not directly related to the current
level of output or input, which are hence assigned to Tt. This break down is illustrated in
9Marketing and Promotion, which is part of the General Services Support Estimate (another measure
parallel to PSE), has also been included in τt(za
i).
11
Figure 2.10
Figure 2 – Agricultural Transfers: τt(za
i)and Tt
Marketing and Promotion
(3%)
(4)
AARI, non-current
(17%)
(1)
Output Based
(28%) (5)
Non-commodity criteria
(5%)
(2)
Input Based
(19%)
(3)
AARI, current
(27%)
The re-partition refers to 2017. Underlined entries denote policies allocated to Tt.
Since 2002, a component of PSE that has risen in importance is counter-cyclical payments
(see Dimitri et al. (2005)). These payments are triggered when agricultural prices fall below
established thresholds and contribute to reducing the volatility of agricultural income. Our
framework does not explicitly model risks that lead to fluctuations in the agricultural price.
Nevertheless, we do not exclude counter-cyclical payments from our measure of transfers,
since they augment farmers’ income and bias labor allocation towards agriculture. Certain
portions of counter-cyclical payments are based on the current level of farm revenue11, and
others are not12. We carefully select entries from the former into τt(za
i)and the latter into
Tt.
Figure 3shows the evolution of Ttas a fraction of total transfers over time. Since the late
1990s, there have been efforts working on removing the dependence of agriculture transfers
on current levels of output (see also Dimitri et al. (2005)). This resulted in a surge of the
10Additional details on our mapping between (τt(za
i), Tt)and the data are provided in the Appendix. A
complete break-down of all policy measures that add up to Producer Support Estimate is also available in
the individual country files OECD webpage.
11For example, crop insurance.
12For example, price loss coverage payments.
12
importance of Ttaround 1997.
Figure 3 – Agricultural Transfers: Ttas a fraction of Total Transfers over time
1980 1990 2000 2010 2020
Year
0
5
10
15
20
25
30
35
Percentage
3.2 Calibration of Model Parameters
We have eight endogenously calibrated parameters (θa, θn, ρ, ¯a, ν, τtH , τtL,Tt), which we jointly
select to match eight moments from the data13 . Table 1provides a summary of the results.
Three parameters θa, θn, ρ govern the cross-sectional distribution of productivities. θa
and θnare the parameters capturing the cross-sectional variance of agricultural and non-
agricultural productivity za
iand zn
i, respectively. ρis a parameter that determines the
correlation between za
iand zn
i. Following Lagakos and Waugh (2013), we discipline these
parameters with three moments: the cross-sectional variances of the non-transitory compo-
nent of income in both the agriculture sector and the rest of the economy, as well as the
ratio of average income between agriculture and non-agriculture sectors.15
13We choose 2006 as the reference year for calibration in order to isolate our model from the 2007-2009
recession. Selecting a different year for calibration make little quantitative changes to our model predictions.
14ρis a parameter in the joint distribution of (za, zn). At our calibrated value of ρ= 0.9, the implied
Spearman rank correlation coefficient is 0.1, suggesting a weakly positive correlation in the productivities.
15The individual-level data on income are retrieved from the March Current Population Survey. This sur-
vey provides information on pre-tax income (salary, business and farm income), without specifying whether
13
Table 1 – Calibration: Parameters and Moments
External Parameters Value Source
Total Factor Productivity {At}Feenstra et al. (2015)
Calibrated Parameters Target Data Model
Variance of za
iθa= 5.8Cross-Sectional Vari. of Ag. Wage 0.14 0.15
Variance of zn
iθn= 2.7Cross-Sectional Vari. of Non-Ag. Wage 0.24 0.22
Correlation in Giρ= 0.914 Ratio of Avg. Wage in Ag. / Non-Ag. 70.1% 68.5%
Consumption Share of ν= 132.3Price-Elasticity of 0.81 0.81
Non-Ag. Output Demand for Ag. Good
Subsistence Level of Agri Cons. ¯a= 0.006 Emp. Share of Ag. Sector in 2006 1.5% 1.6%
Rate of Prop. Subsidy to {τtL}Prop. Subsidy to Bottom 87% Farms 3.5% 3.9%
Bottom 87% Farms (% of Ag. GDP)
Rate of Prop. Subsidy to {τtH }Prop. Subsidy to Top 13% Farms 13.8% 13.0%
Top 13% Farms (% of Ag. GDP)
Lump-Sum Subsidies {Tt}Lump-Sum Subsidies to Ag. GDP 8.0% 8.0%
Proportional Income Tax {tt}Budget Balancing for Gov’t N/A 0.2%
¯aand νenter the utility function u(ca, cn) = ln(ca
i¯a)+νln(cn
i), where ¯ais the subsistence
level of agricultural consumption, and νis the relative preference between agricultural and
non-agricultural consumption. ¯amakes the agricultural good relatively inferior. As the
economy grows, the share of agricultural consumption declines, and so does the share of
employment in agriculture. To calibrate ¯a, we add the 2006 employment share in agriculture
as a target.
νis a parameter that affects the price elasticity of demand for the agricultural good,
which is selected to match the price elasticity in the data. Andreyeva et al. (2010) offers
elasticity estimates for several food categories, and finds values ranging from 0.27 (eggs) to
0.81 (food away from home). We select the highest elasticity from this report, motivated
by the following considerations. The main mechanism at work in our model stems from
a general equilibrium effect through the adjustment in the agricultural price. The more
inelastic demand is, the stronger the general equilibrium effect becomes, and so does our
mechanism. Targeting the upper bound of the elasticities gives us the most conservative
government transfers are included in this measure or not. Agricultural transfers would either enter salary
income directly or through business or farm income indirectly. For this reason, we match our calibration
targets using the after-subsidy income generated by the model.
14
estimate of the general equilibrium effect, and allows us to interpret the welfare gains of our
counterfactual exercises as an upper bound.
τtH , τtL,Ttsummarize agricultural transfers. While θa, θn, ρ, ¯a, ν are assumed to be time
invariant for the horizon of our policy counterfactuals (2001-2017), τtH , τtL, Ttare recalibrated
in each period, so that the model reproduces the evolution of agricultural transfers over these
years. To calibrate τtH , τtL,Tt, we use three policy-related targets: the share of agricultural
subsidies received by the top 13% of farms, the amount of transfers based on the current
level of output, and the amount of transfers independent of current output. We recover the
share of subsidies received by the top 13% from Cai (2019), which reports the distribution
of government payments by farm size. In particular, Cai (2019) shows that the top 13% of
farms are those with 105 or more hectares of land, and these farms receive well over 60% of
agricultural transfers.16
The equilibrium of the calibrated model replicates closely the evolution of the share of
employment in agriculture over the 2000-2017 time horizon (Figure 4) as well as the income
gap between agricultural and non-agricultural sectors (Figure 5).
16For example, in the baseline year of the calibration (2006), we choose τH, τL, T to jointly match the
following moments: (1) lump-sum subsidies (8% of agricultural GDP), (2) proportional subsidies (17% of
agricultural GDP), and (3) fraction of proportional subsidies that is received by the top 13% farms (80%).
Note that (2) and (3) jointly imply that the top 13% of farms receives proportional subsidies that amount
to 17% ×80% = 13.8% of agriculture GDP. Similarly, the bottom 87% of farms receives 17% ×20% = 3.5%.
15
Figure 4 – Agriculture Employment Share: Model vs Data
2000 2005 2010 2015
Year
1.4
1.5
1.6
1.7
1.8
1.9
Percentage
Model
Data
Figure 5 – Agricultural Productivity Gap: Model vs Data
2000 2005 2010 2015
Year
66
68
70
72
74
76
78
80
Percentage
Model
Data
4 Quantitative Analysis
In this section, we use the model calibrated in Section 3to assess the efficiency and redistri-
butional consequences of current agricultural subsidies.
Four sets of counterfactual exercises are conducted. First, we simply eliminate pro-
portional subsidies τt(za
i)and lump-sum subsidies Ttone at a time, and then all together
16
(Subsection 4.1). Second, we design non-distortionary, individual-specific, utility-neutral
income subsidies Tit, in place of τt(za
i)and Tt(Subsection 4.2). Third, we replace τt(za
i)
and Ttwith a budget-neutral consumption subsidy st(Subsection 4.3). Finally, we study a
budget-neutral, uniform lump-sum subsidy to replace the existing policies (Subsection 4.4).
Throughout the subsequent counterfactual policy analysis, we refer to the existing policy
[τt(za
i), Tt]as “Benchmark”. Income of an agricultural worker under the Benchmark policy
is given by
ya
it = (1 tt)
| {z }
General Income Tax
(1 + τt(za
i))
| {z }
Proportional Subsidies
Atpa
tza
i
| {z }
Ag Income before Subsidies
+Tt
|{z}
Lump-Sum Subsidies
.
4.1 Quantitative Exercise I: Removing τt(za
i)and Tt
4.1.1 Removing τt(za
i)
Figure 6displays the time-series of the price of the agricultural good before and after propor-
tional subsidies τt(za
i)are removed.17 The blue solid line represents the agricultural price
pa
t,Bench before removal. The purple dash-dotted line denotes the price augmented by the
proportional subsidy pa
t,Bench (1 + τtL)18, which corresponds to the post-subsidy wage rate
received by farmers. The red dashed line represents the agricultural price after removing
subsidies.
17In this and all subsequent counterfactual exercises, proportional tax rate ttis recalculated to balance
the budget of the government.
18We plot pa
t,Bench (1 + τtL )because τtL is the rate of subsidy received by the majority (bottom 87%) of
farmers.
17
Figure 6 – Removing τt(za
i): Effect on pa
t
2000 2005 2010 2015
Year
0.36
0.38
0.4
0.42
0.44
Relative Price in Agriculture
pa
Bench*[1+ L]
pa
Bench
pa
GE
We analyze the partial and general equilibrium effect of removing τt(za
i)separately. In
partial equilibrium, the direct income loss experienced by agricultural workers is represented
by the shift from the blue to the purple line. Removing τt(za
i)makes it less attractive to
work in the agricultural sector. As a result, fewer agents select into it, reducing the supply
of the agricultural good and pushing up its relative price. This general equilibrium effect
increases agricultural wages and is represented by the shift from the purple to the red line. In
the data, demand for the agricultural good is inelastic. Therefore, removing τt(za
i)triggers
a significant rise in pa
t, which almost fully compensates farmers for the lost proportional
subsidies. This can be observed in Figure 6, where the purple and red lines almost coincide.
This significant general equilibrium effect has direct implications on agricultural em-
ployment, APG, and individual-level welfare changes. Overall, the agricultural employment
share, represented in the left panel of Figure 7, changes little. When τt(za
i)is removed and
pa
tis fixed at the benchmark level, agricultural employment declines significantly from the
blue line to the purple line. In general equilibrium, however, pa
trises and almost fully com-
pensates farmers for the lost subsidies, resulting in much fewer reallocations (the shift from
the purple line to the red line). As a result, the labor distortion associated with proportional
subsidies is relatively small: in any given period, less than 0.5% of the agents in the model
18
are “misallocated” because of them.
Figure 7 – Removing τt(za
i): Effect on Agricultural Employment Share
2000 2005 2010 2015
Year
1
1.2
1.4
1.6
1.8
Agriculture Employment Share
Bench
PE
GE
0 0.5 1 1.5 2 2.5 3
za
0
0.5
1
1.5
2
zn
Non-Agriculture
Agriculture
Bench
GE
PE
The right panel of Figure 7shows the cutoff line for sector selection as a function of
productivities (za
i, zn
i). In the benchmark, agents to the bottom right of the blue line select
into agriculture. The shift of the cutoff line from blue to purple represents the partial
equilibrium change. In general equilibrium, pa
trises and pushes the cutoff line back to the
red one. For all but top levels of za
i, the red line coincides with the blue line, indicating few
exits out of agriculture in general equilibrium. For top levels of za
i, the polygon between the
blue and red lines represents higher-productivity farmers leaving agriculture. It is important
to point out that the distribution of za
ihas a mean of 1.1and standard deviation of 0.3,
which implies that the measure of agents in this polygon is very small.
The agricultural productivity gap also changes little from the removal of τt(za
i), if we
restrict our calculation of agricultural income to the bottom 87% of farmers. Figure 8
illustrates the change. Again, the partial equilibrium effect (blue to purple) is roughly offset
by the general equilibrium effect (purple to red). It is important to notice that Figure 8
only includes the bottom 87% of farmers into the calculation of average agricultural income.
The top 13% of farmers, who receives a proportional subsidy at τH60%, is suffering a
significant loss of income when subsidies are removed. Including these farmers will lead to a
19
widening of APG, as shown in Figure 9.
Figure 8 – Removing τt(za
i): Effect on APG for Bottom 87% of Farmers
2000 2005 2010 2015
Year
50
52
54
56
58
Percentage
Bench
PE
GE
Figure 9 – Removing τt(za
i): Effect on APG for All Farmers
2000 2005 2010 2015
Year
50
55
60
65
70
75
80
Percentage
Bench
PE
GE
Table 2shows the individual-level welfare changes associated with the removal of τt(za
i).
Welfare changes in this and all the following counterfactual policy experiments are measured
in units of consumption equivalents: we calculate the percentage change () in consumption
that is required to bring an agent’s utility from the benchmark to the counterfactual level:
u(1 + ∆) ca
i,Bench,(1 + ∆) cn
i,Bench=uca
i,CF , cn
i,CF .19 The rows and columns in the table
19We also compute the counterfactual changes in income required to keep agents’ utilities at the benchmark
20
correspond to the percentiles of za
iand zn
i, ranging from the 1st (lowest) to the 99th (highest).
For each combination of (za
i, zn
i), we calculate the percentage change in welfare associated
with the counterfactual policy. Agents who select into agriculture are highlighted in bold.
Table 2 – Removing τt(za
i): Percentage Change in Welfare
Partial Equilibrium
zn
i
% 1 5 10 25 50 75 90 95 99
za
i
10.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13
50.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13
10 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13
25 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13
50 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13
75 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13
90 3.12
3.12
3.12 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13
95 3.18
3.18
3.18 3.18
3.18
3.18 3.18
3.18
3.18 0.13 0.13 0.13 0.13 0.13 0.13
99 34.73
34.73
34.73 34.73
34.73
34.73 34.73
34.73
34.73 34.73
34.73
34.73 34.73
34.73
34.73 17.59
17.59
17.59 0.13 0.13 0.13
General Equilibrium
zn
i
% 1 5 10 25 50 75 90 95 99
za
i
1 0.08 0.08 0.08 0.09 0.09 0.09 0.09 0.09 0.1
5 0.08 0.08 0.08 0.09 0.09 0.09 0.09 0.09 0.1
10 0.08 0.08 0.08 0.09 0.09 0.09 0.09 0.09 0.1
25 0.08 0.08 0.08 0.09 0.09 0.09 0.09 0.09 0.1
50 0.08 0.08 0.08 0.09 0.09 0.09 0.09 0.09 0.1
75 0.08 0.08 0.08 0.09 0.09 0.09 0.09 0.09 0.1
90 0.74
0.74
0.74 0.08 0.08 0.09 0.09 0.09 0.09 0.09 0.1
95 0.76
0.76
0.76 0.76
0.76
0.76 0.76
0.76
0.76 0.09 0.09 0.09 0.09 0.09 0.1
99 32.01
32.01
32.01 32.01
32.01
32.01 32.01
32.01
32.01 32.01
32.01
32.01 32.01
32.01
32.01 17.62
17.62
17.62 0.09 0.09 0.1
In partial equilibrium (top panel), removing proportional subsidies hurts farmers directly,
as the negative entries located in the lower left section of the table show. This loss is especially
severe for the top 13% of the farmers, who used to receive a higher rate of proportional
subsidies. In general equilibrium (bottom panel), the rise in pa
tfully compensates the bottom
87% of farmers , while the top 13% of farmers still experiences a welfare loss.
level. Both measures give similar qualitative results. The results for this alternative measure can be found
in the Appendix.
21
4.1.2 Removing Tt
The lump-sum subsidy Ttprovides an equal amount of resources to all farmers, which is
higher for lower-productivity agents relative to their income. Therefore, in comparison to
proportional subsidies τt(za
i),Tthas a stronger redistribution effect.
After Ttis removed, agents have less incentives to select into agriculture, which results
in a decline in the supply of the agricultural good and a surge in the agricultural price pa
t.
The price change is reflected in Figure 10 and the employment change in Figure 11.
Figure 10 – Removing Tt: Effect on pa
t
2000 2005 2010 2015
Year
0.36
0.38
0.4
0.42
0.44
Relative Price in Agriculture
pa
Bench
pa
GE
On the left panel of Figure 11, the agricultural employment share drops from the blue
line to the purple line due to the lost Tt, which is only partially offset through the rise in
pa
t(purple to red line). The gap between the red and blue lines indicates that lump-sum
subsidies are creating significant distortions in the labor allocation.
22
Figure 11 – Removing Tt: Effect on Agricultural Employment Share
2000 2005 2010 2015
Year
1
1.2
1.4
1.6
1.8
Agriculture Employment Share
Bench
PE
GE
0 0.5 1 1.5 2 2.5 3
za
0
0.5
1
1.5
2
zn
Bench
GE
PE
Non-Agriculture
Agriculture
On the right panel of Figure 11, we can observe first a uniform downward shift of the
agriculture selection cutoff from the blue to the purple line, reflecting a reallocation out
of agriculture across all za
iin partial equilibrium. In general equilibrium pa
trises, which
disproportionately benefits farmers with higher productivities. This is demonstrated by the
counter-clockwise rotation of the cutoff from the purple to the red line. Overall, removing Tt
leads agents with lower levels of za
ito move out of agriculture, while some high-productivity
agents join the sector. Quantitatively, we assess that the median and mean productivity
in agriculture increase respectively by 4.4% and 4.0%, which is consistent with the APG
reduction illustrated in Figure 12. APG in partial equilibrium drops from the blue benchmark
line to the purple line. In general equilibrium, however, the rise in APG overshoots the
benchmark level (blue) and settles at the red line, implying that average income in agriculture
is catching up with the rest of the economy by roughly 4% over the considered time period.
23
Figure 12 – Removing Tt: Effect on APG
2000 2005 2010 2015
Year
50
55
60
65
70
75
80
Percentage
Bench
PE
GE
Table 3illustrates the welfare change from the elimination of Tt. In partial equilibrium
(top panel), all farmers lose an income of Tt, which causes higher utility loss for lower-income
farmers, since they are on the steeper portion of their utility function. In general equilibrium
(bottom panel), the rise in pa
tcompensates farmers to an extent that is proportional to their
net agricultural output20. Adding up partial and general equilibrium effects, total change in
income for a farmer is given by
Tt
|{z}
PE
+ ∆pa
t×(ya
ica
i)
| {z }
GE
.
In the bottom panel of Table 3, agricultural workers with lower za
isuffer a net loss of income,
because they are not producing enough net output to offset the lost Tt. Agricultural workers
with higher za
ion the other hand enjoy a net gain.
The welfare impact of removing Ttis not contained to the agricultural sector. In particu-
lar, outside of agriculture, the partial-equilibrium welfare gains (achieved through a lower tt)
are noticeably transformed into losses in general equilibrium. Among agents outside of agri-
culture, low-productivity agents experience the largest welfare loss. This is a consequence
20An agricultural worker both produces and consumes agricultural output. The net output is (ya
ica
i).
24
of the increase in pa
t, combined with Stone-Geary preferences: lower-income agents spend a
larger fraction of their income on the agricultural good, which becomes more expensive. This
points to an alternative rationale for agricultural support policies: they can be interpreted
as a tool to redistribute to the lower-income by keeping the agricultural good cheaper.
Table 3 – Removing Tt: Percentage Change in Welfare
Partial Equilibrium
zn
i
% 1 5 10 25 50 75 90 95 99
za
i
10.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04
5 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04
10 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04
25 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04
50 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04
75 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04
90 -7.76 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04
95 -6.11 -6.11 -6.11 0.04 0.04 0.04 0.04 0.04 0.04
99 -2.52 -2.52 -2.52 -2.52 -2.52 -2.52 0.04 0.04 0.04
General Equilibrium
zn
i
% 1 5 10 25 50 75 90 95 99
za
i
1 -0.03 -0.03 -0.03 -0.02 -0.02 -0.02 -0.01 -0.01 -0.01
5 -0.03 -0.03 -0.03 -0.02 -0.02 -0.02 -0.01 -0.01 -0.01
10 -0.03 -0.03 -0.03 -0.02 -0.02 -0.02 -0.01 -0.01 -0.01
25 -0.03 -0.03 -0.03 -0.02 -0.02 -0.02 -0.01 -0.01 -0.01
50 -0.03 -0.03 -0.03 -0.02 -0.02 -0.02 -0.01 -0.01 -0.01
75 -0.03 -0.03 -0.03 -0.02 -0.02 -0.02 -0.01 -0.01 -0.01
90 -2.4 -0.03 -0.03 -0.02 -0.02 -0.02 -0.01 -0.01 -0.01
95 -0.65 -0.65 -0.65 -0.02 -0.02 -0.02 -0.01 -0.01 -0.01
99 3.15 3.15 3.15 3.15 3.15 3.15 -0.01 -0.01 -0.01
4.1.3 Removing Both τt(za
i)and Tt
When both proportional and lump-sum subsidies are removed, the effect on agricultural
price, employment, APG, and individual welfare combines the results in subsections 4.1.1
and 4.1.2.
From subsection 4.1.1, removing τt(za
i)has limited effect on agricultural price, employ-
ment, and APG (when the top 13% of the farmers is excluded). Therefore, when both τt(za
i)
25
and Ttare removed, the changes in the economy are mainly driven by the elimination of Tt,
as we Figure 13 illustrates.
Figure 13 – Removing both τt(za
i)and Tt: Effect on pa
t, Agricultural Employment Share, and
APG (Excluding Top 13% of Farmers)
2000 2005 2010 2015
Year
0.36
0.38
0.4
0.42
0.44
0.46
pa
Relative Price of Agriculture Output
Bench
GE
2000 2005 2010 2015
Year
0.5
1
1.5
2
Percentage
Agricultural Employment Share
Bench
PE
GE
2000 2005 2010 2015
Year
45
50
55
60
Percentage
Agricultural Productivity Gap
Bench
PE
GE
The welfare changes reported in Table 4show an interesting non-monotonicity. In partial
equilibrium, high-productivity farmers (99th percentile of za
i) and low-productivity farmers
(90th percentile of za
i) experience a larger loss than middle-productivity farmers (95th per-
centile of za
i). This effect stands in general equilibrium, with middle-productivity farmers
switching from a welfare loss to a gain, low-productivity farmers reducing significantly their
loss, and high-productivity farmers still experiencing a significant decrease in welfare. Over-
all, all agents in agriculture benefit from the rise in income generated by the general equilib-
rium effect on pa
t. However, this compensation is not sufficient neither for low-productivity
farmers, who were overwhelmingly benefiting from Tt, nor for high-productivity farmers,
who were receiving a significantly higher rate of proportional subsidy τtH .
26
Table 4 – Removing both τt(za
i)and Tt: Percentage Change in Welfare
Partial Equilibrium
zn
i
% 1 5 10 25 50 75 90 95 99
za
i
1 0.19 0.19 0.19 0.19 0.19 0.19 0.19 0.19 0.19
5 0.19 0.19 0.19 0.19 0.19 0.19 0.19 0.19 0.19
10 0.19 0.19 0.19 0.19 0.19 0.19 0.19 0.19 0.19
25 0.19 0.19 0.19 0.19 0.19 0.19 0.19 0.19 0.19
50 0.19 0.19 0.19 0.19 0.19 0.19 0.19 0.19 0.19
75 0.19 0.19 0.19 0.19 0.19 0.19 0.19 0.19 0.19
90 10.68
10.68
10.68 0.19 0.19 0.19 0.19 0.19 0.19 0.19 0.19
95 9.27
9.27
9.27 9.27
9.27
9.27 9.11
9.11
9.11 0.19 0.19 0.19 0.19 0.19 0.19
99 37.25
37.25
37.25 37.25
37.25
37.25 37.25
37.25
37.25 37.25
37.25
37.25 37.25
37.25
37.25 17.54
17.54
17.54 0.19 0.19 0.19
General Equilibrium
zn
i
% 1 5 10 25 50 75 90 95 99
za
i
1 0.06 0.07 0.07 0.08 0.09 0.1 0.1 0.1 0.11
5 0.06 0.07 0.07 0.08 0.09 0.1 0.1 0.1 0.11
10 0.06 0.07 0.07 0.08 0.09 0.1 0.1 0.1 0.11
25 0.06 0.07 0.07 0.08 0.09 0.1 0.1 0.1 0.11
50 0.06 0.07 0.07 0.08 0.09 0.1 0.1 0.1 0.11
75 0.06 0.07 0.07 0.08 0.09 0.1 0.1 0.1 0.11
90 1.51
1.51
1.51 0.07 0.07 0.08 0.09 0.1 0.1 0.1 0.11
95 0.27 0.27 0.27 0.08 0.09 0.1 0.1 0.1 0.11
99 30.65
30.65
30.65 30.65
30.65
30.65 30.65
30.65
30.65 30.65
30.65
30.65 30.65
30.65
30.65 17.62
17.62
17.62 0.1 0.1 0.11
The welfare gains outside of the agricultural sector show the same type of heterogeneity
displayed when only Ttwas removed. Low-productivity agents experience the smallest gains,
since their consumption is weighted more heavily on the agricultural good, whose price
significantly increases.
4.1.4 Summarizing the Effects of Removing τt(za
i)and Tt
A qualitative comparison of the effects of removing τt(za
i)and Ttis summarized in Table 5.
27
Table 5 – Removing τt(za
i)and Tt: General Equilibrium Effect
Remove τt(za
i)Remove TtRemove Both
Agri Price Increase Increase Increase
Agri Emp Share Little Change Decrease Decrease
APG Widen Shrink Widen
APG (excl. top) Little Change Shrink Shrink
Who Benefits Agents in
non-agriculture
High-productivity
farmers
Agents in non-ag. and
middle-productivity
farmers
Who Loses All farmers, esp.
high-productivity
All agents, excl.
high-productivity
farmers
High-productivity and
low-productivity
farmers
Labor Allocation Little Efficiency Gain Small Efficiency Gain Small Efficiency Gain
Overall, removing τt(za
i)has little effect on agricultural employment, APG, and the
welfare for the bottom 87% of farmers. The top 13% of farmers mechanically suffers a loss
due to the large level of τtH . Removing Ttleads to a decline in agricultural employment and
a shrinking APG. The rise in pa
tis not enough to compensate lower-productivity farmers for
the lost Tt, but it is high enough to make higher-productivity farmers better off. At the same
time, agents outside of agriculture suffer a welfare loss due to the magnitude of the increase
in pa
t, with low-productivity agents experiencing the heftiest loss.
A quantitative comparison is presented in Table 6, with partial equilibrium effects listed
between parentheses. The first row of the table report the percentage changes in the agricul-
tural price pa
tfollowing the counterfactual policy reform. The second row records the changes
in the share of agricultural employment. The third and fourth row report the changes in
APG, both including and excluding the top 13% farmers. A negative change in APG records
a widening of the income gap between the agricultural and non-agricultural sectors.
28
Table 6 – Removing τt(za
i)and Tt: Partial and General Equilibrium Effect
Percentage Remove τt(za
i)Remove TtRemove Both
Avg Min Max Avg Min Max Avg Min Max
%∆pa
t5.9 4.0 11.3 6.8 4.3 10.3 13.6 7.3 22.7
Empa
-0.02 -0.01 0.03 -0.1 -0.3 0.0 -0.1 -0.3 0.0
(-0.5) (-0.9) (-0.3) (-0.6) (-0.9) (-0.3) (-0.9) (-1.4) (-0.5)
APG -9.2 -15.9 -5.4 4.2 2.4 6.3 -6.9 -12.0 -4.3
(-10.4) (-18.4) (-5.8) (-3.8) (-5.7) (-1.9) (-14.2) (-24.2) (-7.6)
APG, Exl 0.1 -0.5 0.9 1.5 0.1 2.6 1.7 0.4 2.8
Top Farmers (-2.4) (-4.1) (-1.4) (-3.9) (-6.3) (-1.9) (-6.3) (-10.3) (-3.3)
% Worse Off 0.2 0.2 0.2 99.2 98.4 99.5 1.0 0.8 1.3
Median Voter Yes Yes Yes No No No Yes Yes Yes
Note: the tables report annual changes calculated from 2000 to 2017; partial equilibrium effects between
parentheses.
The last two rows of Table 6offer an overview of the “political feasibility” of removing
the existing transfers. In particular, the second to last row shows the percentage of agents
in the economy that are made worse off following each counterfactual policy, while the last
row reflects the welfare change of the median voter in the economy21. Two opposing effects
are at play. On one hand, the proportional income taxes ttused to finance either τt(za
i)or
Ttare no longer necessary (“tax-saving” effect), which is welfare-improving for all agents.
On the other hand, removing τt(za
i)or Ttleads to a rise in the agricultural price pa
t(general
equilibrium effect), which is welfare-decreasing for most.
When τt(za
i)is removed, the “tax-saving” effect prevails, and agents outside of agriculture
are better off. This happens because 80% of the proportional subsidies τt(za
i)used to go to
the top 13% of the farmers, with only the remaining 20% going to bottom 87%. Overall,
removing the generous subsidies allocated to high-productivity farmers generates notable
tax savings, while removing small subsidies to bottom 87% of the farmers causes a modest
rise in pa
t(Figure 6). Given that 98.4% of the population is outside the agricultural sector,
only 1.6% (agri emp share) ×13% (top 13% farmers) 0.2% of the population is actually
hurt by removing τa
t(za
i).
21If the median voter experiences a welfare loss, they will vote “No” to a proposal to remove the considered
policy, and “Yes” otherwise.
29
When only Ttis removed, the price pa
tof agricultural goods rises significantly (Figure
10), causing a generalized welfare loss across the economy. The only agents that gain from
this reform are high-productivity farmers, who benefit more than proportionally from an
increase in pa
t, since they also enjoy a high rate of proportional subsidy. Removing this
component of the existing policy, while at the same time maintaining τt(za
i), seems unlikely
to find political support in the economy. When all transfers are eliminated, however, the
“tax-saving” benefits is dominating for an overwhelming majority of agents, suggesting that
such policy has higher potential to find political consensus.
4.2 Quantitative Exercise II: Replacing τt(za
i)and Ttwith Non-
Distortionary Transfers Tit
Employment in agriculture drops after τt(za
i)or Ttis removed. The magnitude of the drop
gives a measure of the labor distortion created by each type of subsidy. This measure,
however, is not welfare-neutral: a direct removal of subsidies benefit certain group of agents
and hurts others, making it difficult to assess the overall welfare impact of reforms.
A welfare-neutral way to assess the efficiency gains is to replace τt(za
i)or Ttwith a scheme
of utility-preserving, non-distortionary transfers. In particular, after removing one form of
agricultural subsidies (proportional or lump-sum), we augment all agents’ income with an
individual-specific transfer or tax Tit. The amount of this transfer/tax is tailored to each
agent so that their utility is preserved at the pre-reform levels.
Since Tit is individual-specific and does not distort sector selection, it provides a more
efficient way of subsidizing the agents in the agricultural sector than the current transfers.
In light of this, we expect the total transfers RTitdG (za
i, zn
i)to generate fiscal savings, for
delivering the same levels of utility for all agents as in the benchmark economy. The total
amount of fiscal saving, therefore, provides an indirect measure of the aggregate efficiency
gain.
The left panel of Figure 14 illustrates the net fiscal revenue from replacing proportional
30
subsidies τt(za
i)with Tit, as a fraction of agricultural GDP. The red dashed line represent
the welfare gain in partial equilibrium, where pa
tis fixed at the benchmark level. The blue
solid line represent the fiscal saving in general equilibrium. Due to general equilibrium
effect, the efficiency gain is much smaller than in partial equilibrium. This indicates a small
distortionary effect of proportional subsidies, which is consistent with our previous analysis
in Figure 7.
Figure 14 – Replacing τt(za
i)or Ttwith Tit: Efficiency Gains
2000 2005 2010 2015
Year
0
2
4
6
8
10
12
Percentage of Agri GDP
GE
PE
2000 2005 2010 2015
Year
0
2
4
6
8
10
12
Percentage of Agri GDP
PE
GE
The right panel of Figure 14 plots the fiscal saving from replacing lump-sum subsidies
Ttwith Tit. Again, the efficiency gains are much smaller in general equilibrium than in
partial equilibrium. The welfare gain from removing proportional subsidies is close the the
gain from removing lump-sum subsidies, both averaging 2% of agricultural GDP over the
considered time period. However, total spending on lump-sum subsidies is only 1/3of that
on proportional subsidies, as shown in Figure 3of Section 3. Therefore, on a per dollar basis,
τt(za
i)is less distortionary than Tt.
Fiscal savings in Figure 14 are also briefly summarized in Table 7below.
31
Table 7 – Removing τt(za
i)and Tt: Partial and General Equilibrium Effect
Percentage of Ag. GDP Remove τt(za
i)Remove Tt
Avg Min Max Avg Min Max
Fiscal Savings (GE) 2.0 0 5.9 2.2 1.0 4.9
Fiscal Savings (PE) 4.7 0.6 12.1 4.9 1.7 10.6
Note: the tables report annual changes calculated from 2000 to 2017; partial equilibrium effect between
parentheses
4.3 Quantitative Exercise III: Replacing τt(za
i)and Ttwith Con-
sumption Subsidy st
In this counterfactual exercise, we analyze a consumption subsidy to the agricultural good,
with the following motivation. As pointed out in Subsection 4.1, one effect of the current
policies is to keep the agricultural good artificially cheap, mostly helping the lower-income,
whose share of consumption in the agricultural good tends to be higher, due to Stone-Geary
preferences. Given this motivation, a consumption subsidy seems a more natural and direct
way of achieving the same policy goal.
The subsidy stis selected to be budget-neutral, in the sense that the total spending on
the consumption subsidy we design is equal to the spending level of current policies. With
this consumption subsidy, an agent’s budget constraint reads
pa
tca
it (1 st) + cn
it =yit.
4.3.1 Replacing τt(za
i)with Consumption Subsidy st
Figure 15 plots the price of the agricultural good before and after streplaces τt(za
i). The
blue line represents the benchmark model. When τt(za
i)is replaced with st,pa
trises to the
red line due to an increased demand for the agricultural good. Despite the fact that the
agricultural good becomes more expensive, the after-subsidy price pa
t(1 st)ends up lower
than the benchmark level (purple line).
32
Figure 15 – Replacing τt(za
i)with st: Effect on pa
t
2000 2005 2010 2015
Year
0.32
0.34
0.36
0.38
0.4
0.42
0.44
Relative Price of Agriculture Output
pa
Bench
pa
ConSub
pa
ConSub*(1-s)
The blue line lies in between the red and purple lines, resulting in two implications.
First, consumers in both sectors are better off due to the drop in after-subsidy price of the
agricultural good (i.e. the purple line is below the blue line). Second, agricultural workers
also benefit because they can sell their output is sold at a higher price (i.e red line is above
the blue line).
The increased demand for the agricultural good also leads to an increase in agricultural
employment, as the left panel of Figure 16 shows. The new employment share under st(red
line) lies around 0.2% higher than the benchmark level (blue line). On the right panel of
Figure 16, increased price pa
tincentivizes more agents to switch into agriculture, especially
those with moderately higher za
i
22.
22The exception is a small mass of agricultural workers with higher productivities za
i, who relocate out
of the agricultural sector after losing proportional subsidies τtH .
33
Figure 16 – Replacing τt(za
i)with st: Effect on Agricultural Employment
2000 2005 2010 2015
Year
1.5
1.6
1.7
1.8
1.9
2
2.1
2.2
Agriculture Employment Share
Bench
Con Sub
0 0.5 1 1.5 2 2.5 3
za
0
0.5
1
1.5
2
zn
Bench
Con Sub
Non-Agriculture
Agriculture
Not all agents in the economy benefit from this policy reform. In fact, the top 13% of
the farmer distribution, who used to receive income subsidies at a rate of τtH 60%, suffer
a net loss. The bottom 87% of farmers are better off. They used to receive income subsidies
at τtL 5%, which are now replaced by a price gain of approximately 10% (given by the
difference between blue and red lines in Figure 15).
The overall welfare impact of this policy reform is summarized in Table 8: welfare im-
proves for all agents, except the top farmers.
Table 8 – Replacing τt(za
i)with st: Percentage Change in Welfare
General Equilibrium
zn
i
% 1 5 10 25 50 75 90 95 99
za
i
1 5.03 5.03 5.02 5.02 5.01 5 4.99 4.99 4.99
5 5.03 5.03 5.02 5.02 5.01 5 4.99 4.99 4.99
10 5.03 5.03 5.02 5.02 5.01 5 4.99 4.99 4.99
25 5.03 5.03 5.02 5.02 5.01 5 4.99 4.99 4.99
50 5.03 5.03 5.02 5.02 5.01 5 4.99 4.99 4.99
75 5.03 5.03 5.02 5.02 5.01 5 4.99 4.99 4.99
90 6.96 5.03 5.02 5.02 5.01 5 4.99 4.99 4.99
95 7.07 7.07 7.07 5.02 5.01 5 4.99 4.99 4.99
99 27.65
27.65
27.65 27.65
27.65
27.65 27.65
27.65
27.65 27.65
27.65
27.65 27.65
27.65
27.65 13.58
13.58
13.58 4.99 4.99 4.99
34
4.3.2 Replacing Ttwith Consumption Subsidy st
As Ttis removed, a fraction of the agricultural workers reallocates outside of agriculture,
as shown in the left panel of Figure 17. This reduces the supply of the agricultural good
and leads to a rise in the agricultural price (Figure 18). On the right panel of Figure 17,
increased price pa
tincentivizes more agents to switch into agriculture, especially those with
moderately higher za
i.
Figure 17 – Replacing Ttwith st: Effect on Agricultural Employment
2000 2005 2010 2015
Year
1.5
1.55
1.6
1.65
1.7
1.75
1.8
1.85
Agriculture Employment Share
Bench
Con Sub
0 0.5 1 1.5 2 2.5 3
za
0
0.5
1
1.5
2
zn
Bench
Con Sub
Non-Agriculture
Agriculture
Figure 18 – Replacing Ttwith st: Effect on pa
t
2000 2005 2010 2015
Year
0.37
0.38
0.39
0.4
0.41
0.42
0.43
0.44
Relative Price of Agriculture Output
pa
Bench
pa
GE
pa
GE*(1-s)
35
The individual-level welfare changes are demonstrated in Table 9. Higher-productivity
farmers enjoy an overall gain from this policy reform, because the rise in price pa
tmore than
compensates them for the lost Tt. On the other hand, every other agent in the economy
suffers a loss, which tends to be higher for low-productivity agents.
Table 9 – Replacing Ttwith st: Percentage Change in Welfare
General Equilibrium
zn
i
% 1 5 10 25 50 75 90 95 99
za
i
1 -0.03 -0.03 -0.03 -0.03 -0.03 -0.02 -0.02 -0.02 -0.02
5 -0.03 -0.03 -0.03 -0.03 -0.03 -0.02 -0.02 -0.02 -0.02
10 -0.03 -0.03 -0.03 -0.03 -0.03 -0.02 -0.02 -0.02 -0.02
25 -0.03 -0.03 -0.03 -0.03 -0.03 -0.02 -0.02 -0.02 -0.02
50 -0.03 -0.03 -0.03 -0.03 -0.03 -0.02 -0.02 -0.02 -0.02
75 -0.03 -0.03 -0.03 -0.03 -0.03 -0.02 -0.02 -0.02 -0.02
90 1.87
1.87
1.87 -0.03 -0.03 -0.03 -0.03 -0.02 -0.02 -0.02 -0.02
95 0.11
0.11
0.11 0.11
0.11
0.11 0.11
0.11
0.11 -0.03 -0.03 -0.02 -0.02 -0.02 -0.02
99 3.7 3.7 3.7 3.7 3.7 3.7 -0.02 -0.02 -0.02
4.3.3 Summarizing the Effects of Replacing τt(za
i)or Ttwith st
A qualitative and a quantitative comparison of the effects of replacing τt(za
i)or Ttwith
consumption subsidy are given in Table 10 and Table 11.
Table 10 – Replacing τt(za
i)and Tt: General Equilibrium Effect
Replace τt(za
i)Replace Tt
pa
tIncrease Increase
pa
t(1 st)Decrease Increase
Agri Emp. Increase Decrease
Who Benefits All Agents except Top 13% Farmers High-Productivity Farmers
Who Loses Top 13% Farmers Low-Productivity Farmers
Table 11 – Replacing τt(za
i)(left) and Tt(right) with st: General Equilibrium Effect
Percentage Replace τ(za
i)Replace T
Avg Min Max Avg Min Max
%∆pa8.8 4.3 16.1 7.5 4.0 12.4
Empa0.2 0.3 0.1 -0.1 -0.2 0.0
Note: the tables report annual changes calculated from 2000 to 2017
36
4.4 Quantitative Exercise IV: Replacing Current Subsidies with a
Lump-Sum Transfer
Another candidate policy to implement the same redistribution effect that the current policy
carries is a budget-neutral, uniform lump-sum subsidy to agricultural workers (which we
identify as a “fair” subsidy).
Practically, this means converting the expenditure allocated to τt(za
i)into Tt, which
be decomposed in two steps. First, we simply remove proportional subsidies τt(za
i), which
generates the same impact already described in Subsection 4.1.1. Second, lump-sum subsidies
are increased, giving agents agents more incentives to select into agriculture, which boosts
agricultural employment (middle panel of Figure 19) and drives down the agricultural price
pa
t(left panel). The additional amount of lump-sum subsidies also induce agents with lower
productivity-pair (za
i, zn
i)to favor agriculture, widening the APG (right panel).
Figure 19 – Converting τt(za
i)into Lump-Sum Subsidies: Effect on pa
t, Agricultural Employ-
ment Share, and APG
2000 2005 2010 2015
Year
0.3
0.32
0.34
0.36
0.38
0.4
0.42
Relative Price of Agriculture Output
pa
Bench
pa
Fair
2000 2005 2010 2015
Year
1.5
2
2.5
Agriculture Employment Share
Benchmark
Fair
2000 2005 2010 2015
Year
50
55
60
65
70
75
80
Percentage
Benchmark
Fair
In terms of welfare, lower-productivity farmers benefit from the new lump-sum subsidies,
which exceed the original proportional subsidies they were receiving. Higher productivity
farmers, who used to enjoy a higher rate of subsidy τtH , experience a loss. Finally, agents
outside agriculture also enjoy a welfare gain, due to the lower price of the agricultural good.
These changes are illustrated in Table 4.4.
37
Table 12 – Convert τt(za
i)into Lump-Sum: Percentage Change in Welfare
General Equilibrium
zn
i
% 1 5 10 25 50 75 90 95 99
za
i
1 5.05 5.04 5.03 5.03 5.02 5.01 5 5 4.99
5 5.05 5.04 5.03 5.03 5.02 5.01 5 5 4.99
10 5.05 5.04 5.03 5.03 5.02 5.01 5 5 4.99
25 5.05 5.04 5.03 5.03 5.02 5.01 5 5 4.99
50 5.05 5.04 5.03 5.03 5.02 5.01 5 5 4.99
75 5.05 5.04 5.03 5.03 5.02 5.01 5 5 4.99
90 7.9 5.04 5.03 5.03 5.02 5.01 5 5 4.99
95 4.85 4.85 4.85 5.03 5.02 5.01 5 5 4.99
99 32.17
32.17
32.17 32.17
32.17
32.17 32.17
32.17
32.17 32.17
32.17
32.17 32.17
32.17
32.17 13.57
13.57
13.57 5 5 4.99
5 Conclusion
In the United States, subsidies and transfers to the agricultural sector are estimated to be
around 10% of agricultural sales in 2018. These sector-specific transfers have the potential of
distorting the allocation of labor across sectors of the economy, by keeping too many workers
in agriculture.
In this paper, we use a general equilibrium model with endogenous sector selection cal-
ibrated to the U.S. economy to assess the welfare consequences of removing and replacing
the existing policies. Four sets of quantitative exercises are conducted.
First, we remove proportional and lump-sum agricultural subsidies one at a time. Both
reforms cause the agricultural price to rise significantly. When proportional subsidies are
removed, there is little impact on agricultural employment, agricultural productivity gap,
and the welfare for the bottom 87% of farmers. The top 13% of farmers suffers a loss due
to the large level of proportional subsidies they are currently receiving. On the other hand,
removing lump-sum subsidies leads to a decline in agricultural employment and a shrinking
agricultural productivity gap. The rise in the agricultural price is not enough to compensate
lower-productivity farmers for the lost lump-sum subsidies (but it is high enough to make
higher-productivity farmers better off). The welfare loss extends to the rest of the economy,
38
with low-productivity agents facing the heftiest cost, due to Stone-Geary preferences.
Second, we measure the efficiency loss from the current subsidies by replacing them with
non-distortionary counterparts. Our non-distortionary transfers are designed to be welfare-
neutral for all agents, allowing us to identify the gain in production efficiency in isolation
from the redistribution effect. We find that both proportional and lump-sum subsidies are
causing an efficiency loss that amounts to around 2% of agricultural GDP. However, since
total spending on lump-sum subsidies is only 1/3of that on proportional subsidies, on a per
dollar basis, proportional subsidies are less distortionary than lump-sum ones.
Finally, we analyze two alternative budget-neutral policies to replace the current sys-
tem: a consumption subsidy, and a uniform lump-sum transfer. When current proportional
subsidies are replaced with consumption subsidies, the agricultural price rises while the after-
subsidy price falls. This indicates a benefit to both agricultural workers and consumers, with
the exception of the top 13% of farmers. The uniform lump-sum transfer achieves a similar
welfare impact, although it leaves high-productivity farmers considerably worse off.
A natural extension of our model is to add a dynamic sector-selection choice, and study
the optimal path of agricultural transfers to balance efficiency and redistribution. The
optimal rate at which subsides should be withdrawn may not be time consistent, as the
social planner may be tempted to raise transfers after agents have taken their costly moving
decision. This dynamic inconsistency may rationalize the large policy transfers to agriculture
persistently observed in most developed nations.
39
References
Daron Acemoglu. Inefficient redistribution. American Political science review, 95(3):649–661,
2001.
Tasso Adamopoulos, Loren Brandt, Jessica Leight, and Diego Restuccia. Misallocation,
selection and productivity: A quantitative analysis with panel data from china. Technical
report, National Bureau of Economic Research, 2017.
Jorge Alvarez. The agricultural wage gap: Evidence from brazilian micro-data. Working
Paper, 2018.
Tatiana Andreyeva, Michael W Long, and Kelly D Brownell. The impact of food prices on
consumption: a systematic review of research on the price elasticity of demand for food.
American journal of public health, 100(2):216–222, 2010.
Francisco J. Buera and Joseph P. Kaboski. Scale and the origins of structural change. Journal
of Economic Theory, 147(2):684 – 712, 2012.
Wenbiao Cai. Technology, policy distortions, and the rise of large farms. International
Economic Review, 60(1):387–411, 2019.
Wenbiao Cai and Manish Pandey. The agricultural productivity gap in europe. Economic
Inquiry, 53(4):1807–1817, 2015.
Francesco Caselli. Accounting for Cross-Country Income Differences. In Philippe Aghion
and Steven Durlauf, editors, Handbook of Economic Growth, volume 1, chapter 9, pages
679–741. August 2005.
Francesco Caselli and Wilbur John Coleman II. The us structural transformation and re-
gional convergence: A reinterpretation. Journal of political Economy, 109(3):584–616,
2001.
40
Chaoran Chen. Untitled land, occupational choice, and agricultural productivity. American
Economic Journal: Macroeconomics, 9(4):91–121, October 2017. doi: 10.1257/mac.
20140171. URL http://www.aeaweb.org/articles?id=10.1257/mac.20140171.
Pavel Ciaian, d’Artis Kancs, and Johan FM Swinnen. Eu land markets and the common
agricultural policy. Working Paper, 2010.
Carolyn Dimitri, Anne B. Effland, and Neilson Conklin. The 20th century transformation
of U.S. agriculture and farm policy. US Department of Agriculture Economic Research
Service, 2005.
Avinash Dixit and John Londregan. Redistributive politics and economic efficiency. The
American Political Science Review, 89(4):856–866, 1995. ISSN 00030554, 15375943.
Robert C. Feenstra, Robert Inklaar, and Marcel P. Timmer. The next generation of the
penn world table. American Economic Review, 105(10):3150–82, October 2015. doi:
10.1257/aer.20130954.
Bruce Gardner. Distortions to agricultural incentives in the united states and canada, 2008.
Douglas Gollin, David Lagakos, and Michael E. Waugh. Agricultural Productivity Differ-
ences across Countries. American Economic Review, 104(5):165–170, May 2014.
Barry K Goodwin, Ashok K Mishra, and François Ortalo-Magné. The buck stops where?
the distribution of agricultural subsidies. Working Paper, 2011.
Berthold Herrendorf and Todd Schoellman. Wages, human capital, and barriers to structural
transformation. American Economic Journal: Macroeconomics, 10(2):1–23, 2018.
Thomas W Hertel. Applied general equilibrium analysis of agricultural and resource policies.
Handbook of agricultural economics, 2:1373–1419, 2002.
Thomas W Hertel and Marinos E Tsigas. Tax policy and us agriculture: a general equilibrium
analysis. American Journal of Agricultural Economics, 70(2):289–302, 1988.
41
Joan Hamory Hicks, Marieke Kleemans, Nicholas Y Li, and Edward Miguel. Reevaluat-
ing agricultural productivity gaps with longitudinal microdata. Working Paper 23253,
National Bureau of Economic Research, March 2017.
Maureen Kilkenny and Sherman Robinson. Economywide implications of agricultural liber-
alization in the united states. (1486-2018-6280):21, 1990. doi: 10.22004/ag.econ.278265.
URL http://ageconsearch.umn.edu/record/278265.
Barrett E Kirwan. The incidence of us agricultural subsidies on farmland rental rates. Journal
of Political Economy, 117(1):138–164, 2009.
David Lagakos and Michael E. Waugh. Selection, agriculture, and cross-country productivity
differences. American Economic Review, 103(2):948–80, April 2013. doi: 10.1257/aer.103.
2.948.
Kiminori Matsuyama. Structural change. In The new Palgrave dictionary of economics.
Palgrave-Macmillan, 2008.
OECD. Agricultural Policy Monitoring and Evaluation 2018. 2018.
Diego Restuccia, Dennis Tao Yang, and Xiaodong Zhu. Agriculture and aggregate pro-
ductivity: A quantitative cross-country analysis. Journal of Monetary Economics, 55(2):
234–250, March 2008.
A. D. Roy. Some thoughts on the distribution of earnings. Oxford Economic Papers, 3(2):
135–146, 1951.
Robert W. Staiger and Guido Tabellini. Discretionary trade policy and excessive protection.
The American Economic Review, 77(5):823–837, 1987. ISSN 00028282.
Marc Teignier. The role of trade in structural transformation. Journal of Development
Economics, 130:45–65, 2018.
42
Appendix
A From PSE to τ(za
i)and T
Data on Agricultural support is retrieved from the OECD (2018). The main policy indi-
cator we use is Producer Support Estimate (PSE). PSE is comprised of a variety of policy
measures in support of the agricultural sector, for which a detailed breakdown is offered by
the OECD. We build our measures of distortion, characterized by τt(za
i)∈ {τtL, τtH }, Tt, by
identifying two distinct sets of policy measures (policy-set aand b) from the data. The first
contains support policies that directly distort the allocation of productive resources between
agriculture and the rest of the economy (for example, commodity price support and input
subsidies). The second is comprised of lump-sum payments and subsidies that are not tied
to current production. Even if these payments do not distort labor demand by agricultural
firms, they still play a role in the agents’ sector choice, since agents do not have access to
these transfers when they chose the non-agricultural sector . In particular, there are 7 main
groups of policies:
1. Support based on commodity outputs
2. Payments based on input use
3. Payments based on current Area/Animals/Receipts/Income, production required
4. Payments based on non-current Area/Animals/Receipts/Income, production required
5. Payments based on non-current A/An/R/I, production not required
6. Payments based on non-commodity criteria
7. Miscellaneous payments
(1) Support based on commodity outputs is divided in 2 sub-groups:
43
Market Price Support
Payments based on output
Market Price Support is defined by the OECD as “an indicator of the annual monetary value
of gross transfers from consumers and taxpayers to agricultural producers arising from policy
measures creating a gap between domestic producer prices and reference prices of a specific
agricultural commodity measured at the farm-gate level. This type of transfers create a
gap between the price received by agricultural producers and the reference price, which we
interpret as the actual market clearing price. By creating this gap, Market Price Support
provides incentives to agricultural firms to hire inputs in excess of what they would in the
benchmark undistorted economy, and for this reason we include it into policy-set a.
Payments based on output account for a variety of polices that help farmers managing
the possible mismatch between harvesting date and sales date. For example, the United
States Department of Agriculture reports that Marketing Assistance Loans23 “provide pro-
ducers interim financing at harvest time to meet cash flow needs without having to sell their
commodities when market prices are typically at harvest-time lows. Allowing producers to
store production at harvest facilitates more orderly marketing of commodities throughout the
year.These payments seem to provide a form of income smoothing to farmers, as well as
insurance against short-term fluctuation in commodity prices. While our model is not really
suited to capture these short-term fluctuations, these transfers are sector-specific and there
is no such transfer outside of agriculture. Since we want to give our model the best chance
to generate sizable distortions, we decided to include these policies in set (a).
(2) Payments based on input use, which are included in policy-set a, collect most of sub-
sidies to production inputs in agriculture, for example Energy subsidies, Grazing Subsidies,
Conservation loans. All these payments reduce the opportunity cost of running an agricul-
tural business, and naturally lead to an allocation of excess resources to the agricultural
23Other fairly similar programs included in Payments based on output are Loan deficiency payments,
Commodity Loan Forfeit, Storage Payments, Commodity loans interest subsidies.
44
sector, compared to a benchmark economy without transfers.
(3) Payments based on current Area/Animals/Receipts/Income, production required in-
clude crop disaster payments and crop insurance payments. We applied the same logic as
Payments based on output: even if our model is not designed to account for the impact of
natural disasters on agricultural production, these programs are sector specific and make the
agricultural sector a more attractive choice for economic agents. Payments based on current
Area/Animals/Receipts/Income, production required, includes also Income Tax Concessions
to the agricultural sector, which reduce the opportunity cost to operate in agriculture. We
include them all in policy-set a.
Both (4) Payments based on non-current Area/Animals/Receipts/Income, production re-
quired and (5) Payments based on non-current A/An/R/I, production not required, are based
on non-current levels of inputs. They don’t influence the current allocation of resources, and
for this reason they are excluded from policy-set a. Nonetheless, they contribute to policy-set
b, with the exclusion of payments of counter-cyclical nature.
(6) Payments based on non-commodity criteria are almost entirely from the Conservation
Reserve Program (CRP), which is a land conservation program administered by the USDA’s
Farm Service Agency that provides payments to farmers to remove environmentally sensitive
land from agricultural production. We include these payments in the policy-set b.
The specific nature of (7) Miscellaneous payments is not reported by the OECD, so they
are also excluded from our computations. Finally, we also add to policy-set athe public
expenditure on marketing and promotion of agricultural goods, which is part of the general
services support to agriculture.
B Additional Figures
Time series of the calibrated rates of proportional subsidies τtL and τtH :
45
Figure 20 – Time Series of τtL and τtH
2000 2005 2010 2015
Year
0
20
40
60
80
100
120
140
Percentage
H
L
B.1 Quantitative Exercise I
Change in income after removing proportional subsidies:
Figure 21 – Removing τt(za
i): impact on income in partial and general equilibrium
Change in income after removing lump-sum subsidies:
46
Figure 22 – Removing Tt: impact on income in partial and general equilibrium
Change in consumption-equivalence of welfare after removing lump-sum subsidies:
Figure 23 – Removing Tt: welfare change in partial and general equilibrium
Change in consumption-equivalence of welfare after removing both proportional and
lump-sum subsidies:
47
Figure 24 – Removing τt(za
i)and Tt: welfare change in partial and general equilibrium
B.2 Quantitative Exercise III
Change in income after replacing proportional subsidies with consumption subsidies:
Figure 25 – Replacing Proportional Subsidies with Consumption Subsidies
Change in income after replacing lump-sum subsidies with consumption subsidies:
48
Figure 26 – Replacing Lump-Sum Subsidies with Consumption Subsidies
Change in consumption-equivalence of welfare after replacing proportional subsidies with
consumption subsidies:
Figure 27 – Replacing Proportional Subsidies with Consumption Subsidies
Change in consumption-equivalence of welfare after replacing lump-sum subsidies with
consumption subsidies:
49
Figure 28 – Replacing Proportional Subsidies with Consumption Subsidies
50
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
We document for 13 countries ranging from rich (Canada, United States) to poor (India, Indonesia) that average wages are considerably lower in agriculture than in the other sectors. Moreover, agriculture has less educated workers and lower Mincer returns. We view these findings through the lens of a multi-sector model in which workers differ in observed and unobserved characteristics and sectors differ in their human-capital intensities. We derive expressions for the implied barriers to the reallocation of labor out of agriculture. We find that in our sample these barriers are considerably smaller than what the macro-development literature has argued.
Article
Recent research has pointed to large gaps in labor productivity between the agricultural and non-agricultural sectors in low-income countries, as well as between workers in rural and urban areas. Most estimates are based on national accounts or repeated cross-sections of microsurvey data, and as a result typically struggle to account for individual selection between sectors. This paper uses long-run individual-level panel data from two low-income countries (Indonesia and Kenya) to explore these gaps. Accounting for individual fixed effects leads to much smaller estimated productivity gains from moving into the non-agricultural sector (or urban areas), reducing estimated gaps by roughly 67%–92%. Furthermore, gaps do not emerge up to 5 years after a move between sectors. We evaluate whether these findings imply a re-assessment of the conventional wisdom regarding sectoral gaps, discuss how to reconcile them with existing cross-sectional estimates, and consider implications for the desirability of sectoral reallocation of labor.
Article
Between 1900 and 2002, mean farm size in the U.S. tripled; productive resources were increasingly concentrated in large farms. These observations are difficult to explain as results of profit maximization by farmers in a frictionless equilibrium model, given exogenous factor endowment and technology. I show that notable shifts in the farm size distribution coincided with important changes in farm legislation, and that farm programs provided larger farms more subsidy per dollar of output produced. These farm‐level distortions are crucial for the increasing dominance of large farms, but have little impact on employment and productivity in agriculture. This article is protected by copyright. All rights reserved
Article
Low agriculture productivity is considered a key obstacle to economic development for many countries. International trade in agricultural goods can help overcome this barrier and facilitate structural transformation because it allows countries to import part of their food needs. This article quantifies the role of trade in this context through the examples of South Korea during the last 50 years and Great Britain in the 19th century. To do the analysis, I calibrate and simulate a two-sector, neoclassical growth model to match the data and perform the policy experiments. I find that agricultural imports played a crucial role in the early transformation of Great Britain, while, in South Korea, trade also had a positive impact on its structural transformation but it could have played a much larger role if the country had not introduced agricultural protection policies.
Article
For 15 European countries over the 1970–2004 period we find large and persistent agricultural productivity gaps, the ratio of value added per hour in nonagriculture to that in agriculture. Comparing the gap in value added per hour to the wage gap between the two sectors suggests that value added in the data is mismeasured. We further find that, controlling for differences in gross domestic product per capita and institutions, the mismeasurement is positively related to self-employed share of hours in agriculture. Correcting for underreporting of self-employment income using our preferred correction factor reduces the measured agricultural productivity gap by 38%. These findings suggest that underreporting can account for a significant portion of the measured agricultural productivity gap. (JEL E01, O47, O52, Q10)
Article
We describe the theory and practice of real GDP comparisons across countries and over time. Version 8 of the Penn World Table expands on previous versions in three respects. First, in addition to comparisons of living standards using components of real GDP on the expenditure side, we provide a measure of productive capacity, called real GDP on the output side. Second, growth rates are benchmarked to multiple years of cross-country price data so they are less sensitive to new benchmark data. Third, data on capital stocks and productivity are (re) introduced. Applications including the Balassa-Samuelson effect and development accounting are discussed.
Article
Recent studies argue that cross-country labor productivity differences are much larger in agriculture than in the aggregate. We reexamine the agricultural productivity data underlying this conclusion using new evidence from disaggregate sources. We find that for the world's staple grains-maize, rice, and wheat-cross-country differences in the quantity of grain produced per worker are enormous according to both micro- and macrosources. Our findings validate the idea that understanding agricultural productivity is at the heart of understanding world income inequality.
Article
This article emloys a computable general equilibrium model to analyze the effects of eliminating farm and food tax preferences in 1977. Tax differentials on capital income, labor payments, and production and sales taxes are each examined. Results indicate that these combined preferences lowered food costs by about $4.5 billion while enhancing after-tax returns to farm land, labor, and capital. The associated general equilibrium tax expenditure is estimated to have been between $5.5 and $6.6 billion.