Land Markets, Resource Allocation, and Agricultural Productivity Chaoran Chen

Diego Restuccia

University of Toronto

University of Toronto and NBER

Ra¨ ul Santaeul`alia-Llopis MOVE, UAB, and Barcelona GSE

May 17, 2017

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

1 / 44

1

Introduction

2

Institutional Background and Data

3

Framework of Analysis

4

The Role of Land Rental Markets

5

Land Rentals and Capital Intensity

6

Robustness

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

2 / 44

Motivation I

Why is agricultural labour productivity so low in poor countries?

I

Important: agriculture accounts for a large portion of rich-poor income differences

I

Factor misallocation is severe in agriculture, related to land institutions

I

We study factor misallocation and its impact on aggregate agricultural productivity using detailed household-level micro data from Ethiopia I

Interesting historical context, state control over land allocations

I

Recent land certification reform to promote tenure security allows us to exploit variation in rentals

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

3 / 44

What We Do

I

Use detailed micro data to measure total factor productivity at the farm level

I

Consider a tractable framework to assess quantitatively I

extent of resource misallocation in agriculture

I

how misallocation affects agricultural productivity

I

whether misallocation is related to land markets across households, locations, and time

I

degree of misallocation within households (across plots), across households, and across crops

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

4 / 44

What We Find I

I

I

Severe factor misallocation: I

Ethiopia agriculture: std(log TFPR) = 1.02

I

An efficient reallocation of resources can increase aggregate agricultural output and productivity by 136%.

Land market matters: I

Overall land rental markets thin (most rentals among relatives and friends) as severe restrictions to rentals remain in place

I

Despite restrictions rentals reduce misallocation across households, locations, and time

I

Reduced misallocation from rentals is associated with the adoption of more capital intensive technologies

Results are robust to exploiting plot-level variation across households and analysis by crops

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

5 / 44

Related Literature

I

Macro development literature on agricultural productivity differences across countries: I

I

Misallocation in agriculture: I

I

Gollin-Parente-Rogerson (2002, 2005), Restuccia-Yang-Zhu (2008), Lagakos-Waugh (2013), Gottlieb-Grobovˇsek (2016)

Adamopoulos-Restuccia (2014), Restuccia-Santaeul` alia-Llopis (2016), Adamopoulos et al (2017), Chen (forthcoming)

Micro development literature studying institutions: I

Acemoglu-Johnson-Robinson (2001), Banerjee-Gertler-Ghatak (2002), Banerjee-Iyer (2005), Goldstein-Udry (2008)

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

6 / 44

1

Introduction

2

Institutional Background and Data

3

Framework of Analysis

4

The Role of Land Rental Markets

5

Land Rentals and Capital Intensity

6

Robustness

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

7 / 44

Why Ethiopia?

I

Growth miracle in Africa: GDP per capita (PPP) growth 6.9% (2005-2015)

I

Agriculture is an important sector: around 75% of employment

I

Detailed micro data to study resource misallocation

I

I

LSMS-ISA data from the World Bank

I

Detailed information on farm-level input and output in physical units

Interesting land institution; land certification reform

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

8 / 44

Institutional Background

I

Current land institution shaped by historical events, but the prevailing characteristic has been state control over the allocation and use of land

I

Property rules evolved from the imperial period (mid 19th century to 1974): I

Allocated land ownership to political supporters regardless of occupation or use in farming, which created a feudal regime

I

Further emergence of private property resulted in powerful landlords

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

9 / 44

Institutional Background I

I

Social injustices lead to a communist regime from 1975 to 1991 “Land to the Tiller”: I

Expropriated all land in the country and redistributed, adjusting for soil quality and family size, among all rural households in the form of use rights

I

Redistributions were frequent, every one to two years, to achieve an equitable allocation of use rights

I

Prohibited land transactions

Since 1991, a market oriented government has largely maintained the policies related to land: I

Ownership still resides with the state

I

Households assigned use rights by village (kebele) and district (woreda) authorities

I

Many of the restrictions to land transactions remain in place

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

10 / 44

Institutional Background—Land Certification I

Land certification reforms have been implemented since early 2000s to mainly promote tenure security by issuing land certificates of use rights

I

The implementation of land reforms were decentralized to local governments: the timing, extent, and enforcement has differed across regions

I

While severe rental restrictions remain in place (e.g. only a fraction of use-rights can be rented, lessee must dwell in the rural area and be engaged only in farming) rental markets have developed at differential pace across regions

I

In 2013, 10.6 percent of agricultural land is rented; 24.7 percent of households formally or informally rent in some land for production

I

The percentage of rented land differs greatly across zones (counties) from 0 to more than 60 percent

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

11 / 44

Data I

The Living Standards Measurement Study – Integrated Surveys on Agriculture (LSMS-ISA) from the World Bank

I

Main data: Ethiopia 2013-2014, 3,629 agricultural households with information on their location I

link households to geographical areas

I

Detailed information on farm-household physical outputs and inputs in agricultural production

I

Each household operates a farm consisting of on average 7 land plots, while typically only one crop is planted in each plot I

I

allows us to decompose productivity within and across households; and across crops

Also explore Ethiopia 2015-2016 for cross-time variation

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

12 / 44

Data I

Output: quantities of each crop produced and price of crops

I

Intermediate inputs: fertilizer and seeds, quantity and price

I

Land inputs: GPS measured field size I

I

I

very detailed information on land quality and rainfall

Capital inputs: I

physical measure of agricultural items and market prices

I

livestocks for agricultural production purposes

Labor inputs: measured in hours I

including family labour, hired labour, and free labour

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

13 / 44

Data

I

Focus on the cropping sector

I

We have detailed measure of land quality: I

Nine elements of land quality: elevation, slope, terrain roughness, nutrient availability, nutrient retention, rooting conditions, excess salts, toxicity, and workability

I

Rain shock: annual precipitation in millimetres

I

We control for these items on output dispersion

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

14 / 44

1

Introduction

2

Institutional Background and Data

3

Framework of Analysis

4

The Role of Land Rental Markets

5

Land Rentals and Capital Intensity

6

Robustness

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

15 / 44

Framework I

Static setting, a single good is produced by a number of heterogeneous production units (farms) I

I

Key: farm productivity directly measured given a production function and data on physical amounts of outputs and inputs

Production function (in per capita form and after controlling for land quality and rain): yi = si1−γ (kiα li1−α )γ , s: ability, y : output, k: capital, l: land

I

si measured from data on {yi , ki , li }: si =

I

yi

1 ! 1−γ

(1−α)γ

kiαγ li

Farm TFP is s 1−γ

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

16 / 44

Framework I

Planner’s problem: maximize output given inputs. X X 1−γ X yi = si (kiα li1−α )γ , s.t. ki = K , max ki ,li

I

I

i

i

i

Efficient allocation: kie =

Psi i

si

K , lie =

i

si

L, and yie =

li = L.

i (

Psi γ (K α L1−α )γ . i si )

The allocations of capital and land are proportional to the farmer’s ability: kie si = , e kj sj

I

Psi

X

lie si = . e lj sj

MPK (MPLa) proportional to APK (APLa) and equalized across farms: MPKi ∝

Chen et al. (2017)

Ye yie = , kie K

∀i.

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

17 / 44

Chen et al. (2017)

10

15 Farm Productivity (log)

20

-10

-15

Land Input (log) -10 -5

Capital Input (log) -5 0 5

0

10

Misallocation in Agriculture

10

15 Farm Productivity (log)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

20

18 / 44

-5

Marginal Product of Land (log) 6 8 10 12 14

Marginal Product of Capital (log) 0 5 10

Misallocation in Agriculture

10

Chen et al. (2017)

15 Farm Productivity (log)

20

10

15 Farm Productivity (log)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

20

19 / 44

Summary Statistics of Misallocation

I

Define TFPRi ≡

yi . kiα li1−α

In efficient allocation:

1−α TFPRi ∝ MPKα ⊥ i. i · MPLi I

I

In our data: std(log TFPR) = 1.02 I

Hsieh and Klenow (2009) manufacturing US 0.49, China 0.63, India 0.67

I

Adamopoulos et al (2017) agriculture China 0.91.

Corr[log(TFPRi ), log(si )] = 0.91: correlated distortion. I

Adamopoulos et al (2017) agriculture China 0.88

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

20 / 44

Summary Statistics of Misallocation

I

Define the efficiency gain as the ratio of efficient output to actual output P d y Ye e = d = Pi ie . Y i yi

I

The overall efficiency gain in agriculture is 2.36-fold

I

Most of this gain (76 percent) acrues by reallocation within zones (counties)

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

21 / 44

1

Introduction

2

Institutional Background and Data

3

Framework of Analysis

4

The Role of Land Rental Markets

5

Land Rentals and Capital Intensity

6

Robustness

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

22 / 44

Farmers Operating Rented Land

I

For each farmer, we have information on whether any cultivated land is rented in

I

Only 24.7% of households formally or informally rent in any land

I

We separate farmers into two groups: with rented land (Di = 1) and without rented land (Di = 0)

I

Separately compute the efficiency gains for each group: P e P e y (Di = 0) Y0e Y1e i yi (Di = 1) , e0 = d = P i id e1 = d = P d Y1 Y0 i yi (Di = 1) i yi (Di = 0)

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

23 / 44

Farmers with and without Rented Land

Whole Sample -

No Rental (0%)

Some Rental (>0%)

S.D. (log TFPR) S.D. (log MPLa)

1.02 0.94

1.05 0.99

0.90 0.78

Efficiency gain

2.36

2.39

2.14

Observations Sample (%)

2,672 100

2,011 75.3

661 24.7

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

24 / 44

Farmers with and without Rented Land

I

Define averages, TFPR =

I

I

Yd K α L1−α

and MPLa =

Yd L

  i Individual level misallocation can be measured by log TFPR and TFPR   i . log MPLa MPLa Regress these measures on Di (=1 if renting land, 0 otherwise). The estimated coefficients are −0.15 and −0.11, both significant at 1% level.

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

25 / 44

Detailed Information on Land Rentals I

Do land rentals help alleviate misallocation?

I

Yes, previous analysis between farmers operating some rental land and no rental land (only land-use right)

I

Who do farmers rent from?

I

Most rentals are between relatives and friends, hence are rentals directing land to best uses?

I

Farmers with rentals are further classified into non-market rentals (whose rental rates are below half of the median rental rates) and market rentals

I

We report the standard deviation of (log) marginal product of land (MPLa) and (log) revenue productivity (TFPR) among farmers of different rental categories, controlling for location fixed effect, farm productivity, and land quality

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

26 / 44

A Detailed Investigation of Land Rentals Standard Deviation (log) (1) (2) (3)

Obs.

MPLa No rental Some rental Non-market rental Some market rental

0.99 0.78 0.79 0.76

0.88 0.71 0.72 0.71

0.48 0.38 0.39 0.37

2,011 661 173 488

TFPR No rental Some rental Non-market rental Some market rental

1.05 0.90 0.94 0.87

0.92 0.81 0.85 0.79

0.38 0.34 0.36 0.33

2,011 661 173 488

Yes

Yes Yes

Yes Yes Yes

– – –

Control variables: Land quality Zone fixed effect Farm productivity Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

27 / 44

Land Rental Markets across Locations I

Four levels of administrative divisions in Ethiopia: regions, zones (counties), woreda (districts), and kebele (wards). We mainly focus on the zone level analysis I

I

I

Reasonable number (56) of zones and enough observations (around 50) within each zone

We exploit the variation of implementation across locations to study the impact of land rentals I

This approach is similar to Banerjee-Iyer (2005)

I

The percentage of land rentals differs from 0 to more than 60% at the zone level, smaller range for market rentals

Re-compute the percentage of rentals as well as measures of misallocation and efficiency gains in each zone

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

28 / 44

.5

Dispersion of Log MPL 1

1.5

Land Rental Markets across Locations

.01

Chen et al. (2017)

.05 .1 .25 Rental Percentage Land Markets, Resource Allocation, and Agricultural Productivity

.5

1

May 17, 2017

29 / 44

Productivity Impact I

We regress measures of misallocation and efficiency gain on zone level rentals: log ez = β0 + β1 log(Rz ) + β2 log TFPz + β3 VzTFP + εz . Dependant Variable Efficiency Gain Std TFPR Std MPLa log Rz

I

-0.048 (0.028)

-0.045 (0.014)

-0.028 (0.018)

One percentage point higher land rental implies 0.8 p.p. efficiency gain I

Since β2 =

∆ez Rz ∆Rz ez

, we can calculate the efficiency gain from rental:

ez 1.85 ∆ez = βˆ1 ∆Rz = −0.054 · 1 · = 0.8 Rz 10.7%

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

30 / 44

Variations of Rentals across Time I

Use two waves of data: Ethiopia 2013-2014 and 2015-2016. Aggregate rentals do not change over this time period. Therefore, we explore household-level variation.

I

Separate households into two groups: those whose market rentals increase over time and otherwise. Then compare the changes over time. Moments

Year

Std. TFPR

2015 2013 Diff

0.909 0.923 -0.014

1.093 0.997 0.096

2015 2013 Diff

0.819 0.853 -0.034

0.924 1.014 0.090

Std. MPL

Chen et al. (2017)

Groups of Households Increase Rentals Otherwise

Land Markets, Resource Allocation, and Agricultural Productivity

Differences

0.110 (0.062)

0.124 (0.063) May 17, 2017

31 / 44

Variations of Rentals across Time I

I I

Consider the following regression ξit = γi + λt + βdi λt + εit ,     MPLai i ξit : household level misallocation log TFPR or log TFPR MPLa di,2015 = 1 if household i increases market rental in 2015; dit = 0 otherwise. Dependent Variable

Dispersion of TFPR (1) (2)

Dispersion of MPLa (3) (4)

β

-0.12 (0.05)

-0.12 (0.05)

Farm FE Zone FE Time FE Observations R2 Chen et al. (2017)

-0.10 (0.04)

Yes

-0.10 (0.04)

Yes

Yes

Yes Yes

Yes

Yes Yes

4,464 0.45

4,464 0.13

4,464 0.54

4,464 0.08

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

32 / 44

Land or Product Market Frictions?

I

I

Extremely poor public infrastructure in Ethiopia, access to markets may be an issue We correlate our measure of farm level distortions with distance to market as a proxy for product market distortions

I

The correlation is weak, around -0.10, although significant (similar for distance to nearest city or major road)

I

Since product market distortions are likely to be common among farmers within a location, consistent with earlier finding that most misallocation within zones

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

33 / 44

Broader Consequences of Misallocation

I

Misallocation—the inability for productive farmers to operate larger farm scales—may discourage the adoption of more advance technologies in farming

I

Land preparation activity conducive to adoption of capital to substitute for labor

I

Investigate role of land rentals on farm capital for land preparation

I

We run a Probit regression on the dummy of household-level land rental, log farm TFP, log farm size, and log TFPR to determine the probability of using capital (tractors or livestocks) and/or human labor in land preparation among households

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

34 / 44

Land Rentals and Land Preparation

I

Cross section data:  TFPR  i Ki∗ = β0 + β1 Di + β2 log(TFPi ) + β3 log(FSi ) + β4 log + εi , TFPR Ki = 1[Ki∗ > 0]. I

β1 = 0.13 (se = 0.07)

I

Consider a farm with average farm size, TFP, and distortions. Rented land makes it 4.4% more likely of using capital in land preparation.

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

35 / 44

Land Rentals and Land Preparation I

Cross-time variation:  TFPR  i kit∗ = β0 + β1 ζit + β2 log(TFPi ) + β3 log(FSi ) + β4 log + εi , TFPR I

ki,2015 = 1 if farm i does not use capital in year 2013/14 while it starts to use capital in year 2015/16 (newly adopted); ki,t = 0 otherwise.

I

ζi,2015 = 1 if farmer i increases his market rentals in year 2015/16 compared to 2013/14; ζi,t = 0 otherwise.

I

I

βˆ1 = 1.41 (se = 0.07) Consider a farm with average farm size, TFP, and distortions. Such farm with more rentals is 32% more likely of starting to use capital in land preparation.

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

36 / 44

1

Introduction

2

Institutional Background and Data

3

Framework of Analysis

4

The Role of Land Rental Markets

5

Land Rentals and Capital Intensity

6

Robustness

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

37 / 44

Plot-level Analysis I

We have information on physical (quantity) of inputs and outputs with a high degree of precision minimizing measurement and specification errors I

Recall that s is computed from si =



yi



1 1−γ

(1−α)γ

kiαγ li

.

Any measurement error in {yi , ki , li } will affect {si } I

Land quality and rainfall shocks are controlled for in our analysis

I

We explore features of our data to further access the robustness of our quantitative results: I

A household on average operates 14 land plots

I

Focus on plot-level analysis instead of household-level and compare results

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

38 / 44

Plot-level Analysis I

Baseline analysis: at the household level, inputs and outputs are aggregated from separate plots.

I

An alternative approach:

I

I

Compute sip on each plot p of household i;

I

Replace the household-level productivity si using the geometric mean, median, max2 of the plot-level productivity sip ;

I

Redo the quantitative analysis and compare the results with our baseline analysis

Also use replace the household-level productivity si using the geometric mean of its 2013 and 2015 levels.

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

39 / 44

Plot-level Analysis

1

√

(Πj sij ) J

Medianj (sij )

max 2j {sij }

Rank Corr with si

0.72

0.71

0.57

0.71

Efficiency Gain

2.59

2.60

2.67

2.33

Observations

2,672

2,672

2,672

2,672

I

Baseline efficiency gain: 2.36-fold

I

Quantitative results robust to different approaches

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

si,2013 si,2015

May 17, 2017

40 / 44

Land Quality Differences by Plot

I

Asses whether rented land has higher quality than non-rented land at the plot level

I

Construct the land quality index: I

Regress the log value added on variables of land quality

I

Construct a quality index q based on variables of land quality

I

Rented land has on average only 3% higher quality than other land plots

I

The causality is not identified

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

41 / 44

Misallocation by Crop I

Plot level analysis also allows us to study misallocation within crop production since most plots produce only one crop

I

We estimate productivity s for each household and each crop. Then we calculate the efficiency gain within each crop

Crop Maize Sorghum Tea Leaves Coffee Wheat Barley Horse Beans

Chen et al. (2017)

Efficiency Gain

Farms (%)

Land (%)

2.48 2.31 1.94 5.98 2.25 2.37 3.79

54.0 41.7 38.3 27.8 23.9 22.5 21.0

13.6 17.2 5.5 7.8 2.7 2.5 2.2

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

42 / 44

Conclusions

I

Misallocation is severe in Ethiopia’s agricultural sector; reallocating resources to best use can increase TFP by 2.36-fold

I

Misallocation connected to land institutions

I

Despite certification reforms, severe restrictions to rentals remain in place with most rentals among relatives and friends

I

Nevertheless, market rentals significantly reduce misallocation across households, locations, and time

I

Across locations, a one percentage higher land rental is associated with a reduction of 0.8 p.p. efficiency gain

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

43 / 44

Conclusions

I

Evidence that reduced misallocation increases the adoption of better technologies in agriculture

I

Measurement errors of inputs and outputs are unlikely to drive our results

I

Misallocation tends to be more severe within cash crops compared to food crops

Chen et al. (2017)

Land Markets, Resource Allocation, and Agricultural Productivity

May 17, 2017

44 / 44