Agricultural Productivity Differences Across Countries Douglas Gollin Oxford

David Lagakos UCSD

Michael E. Waugh NYU

January 14, 2014

0/1

Cross-Country Productivity Differences

• Well known that aggregate Y /N differences are large

• Recent literature: differences even larger in agriculture

• Caselli, 2005; Restuccia, Yang Zhu, 2008

1/1

Value Added Per Worker at International Prices, 1985

Ratio of 90th to 10th percentile country Aggregate

22

Agriculture

45 Source: Caselli (2005)

2/1

Are Agricultural Productivity Differences Really So Large?

• Data for agriculture come from FAO via national sources

• Host of measurement concerns • Exclude home production, which is very common in poor countries?

• Agricultural output data otherwise badly measured?

• Bias induced by international prices?

3/1

Are Agricultural Productivity Differences Really So Large?

Given how shaky these data on agricultural production are, it is surprising how many studies lend credibility to the FAO data series and make bold statements on long term trends and short-run fluctuations. (Jerven, 2013, pg 78)

4/1

This Paper • Look at direct productivity measures for staple grains

• Specifically: Maize, Rice and Wheat

• Combine for roughly half of all calorie consumption

• Lots of micro estimates on output and inputs

5/1

This Paper • Very large Y /N differences across countries for grains

• Find no disagreement between macro (FAO) and micro sources

• Grain Y /N differences similar size as for agricultural sector

• Suggest agricultural Y /N differences NOT just measurement error

6/1

Agricultural Productivity Across Countries

7/1

Agricultural Productivity Data from FAO

• Numerator: gross agricultural output at international prices

- at common set of prices of agricultural goods - no adjustment for intermediate usage • Denominator: economically active population in agriculture

• For 2007 or closest available year

8/1

Agricultural Productivity Differences Across Countries

Ratio Richest Ten Percent over Poorest Ten Percent

50.1

Richest Quarter over Poorest Quarter

29.9

Source: authors’ calculations using data from FAO

9/1

Decomposing Agricultural Output per Worker

• Output per worker in agriculture:

output land output = × worker land worker • output/land - i.e. ”yield,” widely measured in micro studies

• land/worker - also widely measured in micro studies

If output/worker mismeasured, culprit must be one of these two

10 / 1

Decomposing Agricultural Output per Worker

• Output per worker in agriculture:

output land output = × worker land worker • output/land - i.e. ”yield,” widely measured in micro studies

• land/worker - also widely measured in micro studies

If output/worker mismeasured, culprit must be one of these two

10 / 1

FAO Estimates of Output per Land

Table : Tons Produced per Hectare of Staple Grains by Country Group

Maize

Rice

Wheat

Richest Ten Percent of Countries

9.2

8.1

4.9

Richest Quarter

8.2

6.8

4.8

Poorest Quarter

2.4

3.2

2.1

Poorest Ten Percent of Countries

2.0

2.9

2.0

Ratio of Top to Bottom Ten Percent

4.7

2.8

2.5

Ratio of Top to Bottom Quarter

3.4

2.1

2.2

11 / 1

Micro Estimates of Output per Land

• We compiled many independent estimates

• Many from natural sciences, e.g. Field Crop Research

• Others from microeconomic studies

• All: tons produced per hectare for maize, rice or wheat

12 / 1

1

Tons Per Hectare (FAO) 3 5 7

9 111315

Output per Land: FAO versus Micro Estimates

NLD00 UK10 USA00 CHN98 CHN06

ZWE08

UGA08 ZAF03

PHL06 IND10 USA00 IND10 IND10

USA00 USA06 USA11

BNG00

KEN08 CIV08 GHA08 MWI00 AUS00 GHN06 KEN02 TAN08 KEN00 BEN08 NIG02 TOG08

KEN04

MWI90 MOZ08

1

3 5 Tons Per Hectare (Micro Studies)

7

9

11 13 15

13 / 1

Output per Land: FAO versus Micro Estimates

• No real disagreement here

• Richest countries produce at least twice as much grain per hectare

as poorest • What about land per worker?

14 / 1

Output per Land: FAO versus Micro Estimates

• No real disagreement here

• Richest countries produce at least twice as much grain per hectare

as poorest • What about land per worker?

14 / 1

FAO Estimates of Land per Worker

Hectares per Worker Richest Ten Percent of Countries

44.6

Top Quarter

23.9

Bottom Quarter

1.3

Poorest Ten Percent of Countries

1.4

Ratio of Top to Bottom Ten Percent

31.2

Ratio of Top to Bottom Quarter

18.0

15 / 1

Micro Estimates of Land Per Worker

• We compiled independent estimates from household surveys

• Living Standards Measurement Surveys (LSMS)

• Measure: hectares per worker; rural households

• Surveys are for developing countries

16 / 1

Land per Worker: Some Micro Estimates from LSMS • Malawi, 2010: 0.4 hectares / worker

• Ethiopia, 2011: 0.3 hectares / worker

• Tanzania, 2010: 0.3 hectares / worker

• Guatemala, 2003: 1.9 hectares / worker

• Nigeria, 2010: 0.5 hectares / worker

17 / 1

Conclusions

• Direct measures of Y /N in staple grains ⇒ very large differences

across countries

• Macro and micro data sources agree about this

• Suggest large agricultural Y /N differences across countries are real

18 / 1