off-farm income and smallholder commercialization: evidence from ethiopian rural household panel...

23
Off-farm Income and Smallholder Commercialization: Evidence from Ethiopian Rural Household Panel Data By Tesfaye B. Woldeyohanes

Upload: gyles-walters

Post on 20-Jan-2016

226 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Off-farm Income and Smallholder Commercialization: Evidence from Ethiopian Rural Household Panel Data By Tesfaye B. Woldeyohanes

Off-farm Income and Smallholder Commercialization:

Evidence from Ethiopian Rural Household Panel Data

ByTesfaye B. Woldeyohanes

Page 2: Off-farm Income and Smallholder Commercialization: Evidence from Ethiopian Rural Household Panel Data By Tesfaye B. Woldeyohanes

2

Content

• General Background• Research problem • objectives• Hypothesis• Methodology• Results

Page 3: Off-farm Income and Smallholder Commercialization: Evidence from Ethiopian Rural Household Panel Data By Tesfaye B. Woldeyohanes

3

General Background

• Agricultural sector is essential for overall economic transformation of low income countries (World Bank 2007)

• Transforming the sector from subsistence to more marketed oriented production system i.e smallholder commercialization

• Smallholder commercialization is not only supplying surplus product to market (Von Braun et.al. 1994; Pingali and Rosegrant 1995)

• It looks at both the output and input side of production

General Background

Data and Emperical Models

Objectves

Hypothesis

Results

Problem statement

Page 4: Off-farm Income and Smallholder Commercialization: Evidence from Ethiopian Rural Household Panel Data By Tesfaye B. Woldeyohanes

4

Smallholder commercialization- concepts

• On the output side – could be possible with cash crop

or staple food crops

• Input side – taded and owned inputs are valued at

market price.

• It is a process which passes through subsistance, semi-

commercial and fully commercial phases (Pingali and

Rosegrant 1995)

• So, smallholder commercialization passes through these

phases and may not imply immedaite move on to

production of high value cash crops.

General Background

Data and Emperical Models

Objectves

Hypothesis

Results

Problem statement

Page 5: Off-farm Income and Smallholder Commercialization: Evidence from Ethiopian Rural Household Panel Data By Tesfaye B. Woldeyohanes

5

Smallholder commercialization- concepts

• Production of marketable surplus of staple food crops over what is required for household consumption .

• smallholder farmers are constrained by a numerous factors to participate in exchange economy and materialize its potential welfare gains.

General Background

Data and Emperical Models

Objectves

Hypothesis

Results

Problem statement

Page 6: Off-farm Income and Smallholder Commercialization: Evidence from Ethiopian Rural Household Panel Data By Tesfaye B. Woldeyohanes

6

Problem statement

• In Ehiopia, smallholder farmers produce 90% of

total agricultural production on average land

holding of less than 1 ha per HH (CSA 2011)

• They are highly subsistance oriented

• Policy intervention – to promote smallholder

commercialization over the past two decades

• Information dearth – if there is induced behavioral

change in terms of market participation and factors

that determine degree of commercialization

General Background

Data and Emperical Models

Objectves

Hypothesis

Results

Problem statement

Page 7: Off-farm Income and Smallholder Commercialization: Evidence from Ethiopian Rural Household Panel Data By Tesfaye B. Woldeyohanes

7

• Previous studies are

regional (Woldehanna 2000)

focus on few crops (Gebreselassie and Sharp 2007)

or followed project intervention (Geberemedhin and

Jaleta 2010)

relied on cross section analysis

• Is off-farm income help commercialization or slow

down the process?

• Results of this study will help to better understand

the situation, explore policy options and rationally

address it

General Background

Data and Emperical Models

Objectves

Hypothesis

Results

Problem statement

Page 8: Off-farm Income and Smallholder Commercialization: Evidence from Ethiopian Rural Household Panel Data By Tesfaye B. Woldeyohanes

8

Objectives

• To assess trends of output market participation and degree of commercialization of smallholder farmers

• To identify factors that explain the difference in degree of commercialization among households, with specific attention on the role of off-farm income.

General Background

Data and Emperical Models

Objectves

Hypothesis

Results

Problem statement

Page 9: Off-farm Income and Smallholder Commercialization: Evidence from Ethiopian Rural Household Panel Data By Tesfaye B. Woldeyohanes

9

Hypothesis

• The main interest is to test if off-farm income enhances smallholder commercialization in Ethiopia

• Off-farm income can have either negative or positive impact on household degree of commercializtion

General Background

Data and Emperical Models

Objectves

Hypothesis

Results

Problem statement

Page 10: Off-farm Income and Smallholder Commercialization: Evidence from Ethiopian Rural Household Panel Data By Tesfaye B. Woldeyohanes

10

Data and Emperical ModelsData• ERHS data is a unique longitudinal data conducted

in seven rounds from 1989 to 2009• Farming systems were considered as an important

stratification basis in selecting villages.• Based on main agro-ecological zone and sub-zones

1-3 villages were selcted per strata.• About 1477 randomly selected households were

included in 1994 round and re-interviewed in 1995, 1997, 1999, 2004 and 2009.

• The households are from 15 rural villages of 4 major regions.

General Background

Data and Emperical Models

Objectves

Hypothesis

Results

Problem statement

Page 11: Off-farm Income and Smallholder Commercialization: Evidence from Ethiopian Rural Household Panel Data By Tesfaye B. Woldeyohanes

11

Data and Emperical Models

• In this study, I use three waves of data (1997, 1999 and 2004).

• I have a balanced panel data observations for 1184 farm households

• Data on input and output price were collected at the community level during each survey year

• 2004 price is used as constant price for all rounds

General Background

Data and Emperical Models

Objectves

Hypothesis

Results

Problem statement

Page 12: Off-farm Income and Smallholder Commercialization: Evidence from Ethiopian Rural Household Panel Data By Tesfaye B. Woldeyohanes

12

Emperical Model specification

• The underlying emperical model is specified as;

(1)

Where - is the household i HCI or value of crop sold to market in period t• Household commercialization Index (HCI) by Govereh et al

(1999) and Strasberg et al (1999) is the most commonly used indicator of degree of commercialization.

General Background

Data and Emperical Models

Objectves

Hypothesis

Results

Problem statement

Page 13: Off-farm Income and Smallholder Commercialization: Evidence from Ethiopian Rural Household Panel Data By Tesfaye B. Woldeyohanes

13

Emperical Model specification

(1)

- is the household i off-farm income earned in period t - is the household i total value of crop produced in period t - is a vector of household characteristics and resource endowments period t - is a vector of household and village level market access variables - is a regional dummy to capture the difference in terms of agro-ecology between regions , , and are the corresponding parameters to be estimated• The model has two error components - time invariant ind. household heterogeneity, assumed to be normally distributed with mean zero and variance - idiosyncratic error term

General Background

Data and Emperical Models

Objectves

Hypothesis

Results

Problem statement

Page 14: Off-farm Income and Smallholder Commercialization: Evidence from Ethiopian Rural Household Panel Data By Tesfaye B. Woldeyohanes

14

Table 1:Descriptive summary of variables used in estimations (panel)

General Background

Data and Emperical Models

Objectves

Hypothesis

Results

Problem statement

Variables observations mean Std. dev. min max

Market participation * 3552 0.678 0.467 0 1

Value of crop sold (ETB) 3552 1020.491 4347.659 0 189607.600

HCI (%) 3552 32.238 30.0152 0 100

Off-farm participation * 3552 0.417 0.493 0 1

Off-farm income (ETB) 3552 202.330 1022.945 0 47027.78

Age of household head (year) 3552 48.616 15.203 15 105

Male household head* 3552 0.744 0.437 0 1

Literate household head* 3552 0.266 0.442 0 1

Family size (no) 3552 5.711 2.629 1 26

Available family labor (person) 3552 2.931 1.532 0.200 16.200

Farm land size owned (ha) 3552 1.469 1.262 0.050 9.877

Value of crop produced (ETB) 3552 2170.002 5874.020 0 255354.800

Equine owned (no) 3552 0.817 1.468 0 24.000

Livestock owned (TLU) 3552 3.051 3.244 0 58.300

Distance to the nearest market

(km)

3552 10.663 5.810 1 25.000

Involvement in extension

program *

3552 0.108 0.310 0 1

Page 15: Off-farm Income and Smallholder Commercialization: Evidence from Ethiopian Rural Household Panel Data By Tesfaye B. Woldeyohanes

15

Model selection• Modeling smallholder market participation can be a

bit tricky because not all farm households sell their crop in market

• Excluding non-participation (zero values of crop sold and HCI) from the sample may lead to sample selection bias and biased regression parameters.

• 32% of sample households did not participate in crop market as seller, so it is not appropriate to estimate the linear model specified in equation (1)

• In cross section contex, most often Tobit model, Sample selection model or its variant like “Two tier” or “double Hurdle” are used.

General Background

Data and Emperical Models

Objectves

Hypothesis

Results

Problem statement

Page 16: Off-farm Income and Smallholder Commercialization: Evidence from Ethiopian Rural Household Panel Data By Tesfaye B. Woldeyohanes

16

Model selection• Sample selection and double hurdle model include

two steps reflecting the dual decision making process:

Decision 1: Whether to participate in marketDecision 2: How much to sell (volume of

transaction)

• Tobit model assumes the same set of variables and parameters determine both the probability of market participation and the volume of transaction

• zero values associated with non-participation are outcome of rational choice i.e. corner solution

• Somewhat restrictive assumptions

General Background

Data and Emperical Models

Objectves

Hypothesis

Results

Problem statement

Page 17: Off-farm Income and Smallholder Commercialization: Evidence from Ethiopian Rural Household Panel Data By Tesfaye B. Woldeyohanes

17

Model selection

• In this study, we assume variables that determine the household’s decision to participate in output market are also the ones responsible to determine the volume of transaction. So, the tobit model is specified as follows;

(2)

(3)

(4)

General Background

Data and Emperical Models

Objectves

Hypothesis

Results

Problem statement

Page 18: Off-farm Income and Smallholder Commercialization: Evidence from Ethiopian Rural Household Panel Data By Tesfaye B. Woldeyohanes

18

Model selection

Where : is the latent variable which is the desired (or potential) market participation by households is the actual observed market participation

) and )

• My objective is to obtain consistent estimate of the parameter .

• Unlike cross section data context, panel data tobit model introduces individual effect, that complicates estimation of parameters of interest

General Background

Data and Emperical Models

Objectves

Hypothesis

Results

Problem statement

Page 19: Off-farm Income and Smallholder Commercialization: Evidence from Ethiopian Rural Household Panel Data By Tesfaye B. Woldeyohanes

19

Model selection

• In this study, Random Effect tobit model is estimated because the time series is short and we have substantial time invariant regressors in our model.

• However, RET makes strong assumptions like the individual effect is normally distributed and uncorrelated with regressors.

• FE could relax this assumption, however it is impossible to estimate the parameters independent of individual effects in panel data context.

General Background

Data and Emperical Models

Objectves

Hypothesis

Results

Problem statement

Page 20: Off-farm Income and Smallholder Commercialization: Evidence from Ethiopian Rural Household Panel Data By Tesfaye B. Woldeyohanes

20

Model selection

• The other option is to estimate Pooled tobit model, which relaxes the strict exogeneity of regressors in RET model.

• This estimation approach also produces consistent (though inefficient) estimate as noted in Maddala (1987)

• The estimation is done by maximum likelihood estimation technique

• The conditional and unconditional marginal effect on actual volume transacted and HCI per unit change in the explanatory variables is calculated.

General Background

Data and Emperical Models

Objectves

Hypothesis

Results

Problem statement

Page 21: Off-farm Income and Smallholder Commercialization: Evidence from Ethiopian Rural Household Panel Data By Tesfaye B. Woldeyohanes

21

Preliminary ResultsTable 2: Summary descriptive statistics for HH characteristics and market participation by year (n=1184) General

Background

Data and Emperical Models

Objectves

Hypothesis

Results

Problem statement

Variables 1997 1999

2004

mean SD mean SD Mean SD

Value of crop produced 2108.854 3216.014 1798.395 2097.275 2602.758 9407.443

Market participation 0.661 0.473 0.646 0.478 0.726 0.446

Value of crop sold (ETB) 950.753 2473.315 814.512 2473.315 1296.207 6933.838

HCI (%) 29.371 27.948 32.147 31.781 35.196 29.935

Off-farm participation 0.290 0.454 0.519 0.500 0.440 0.497

Off-farm income (ETB) 422.245 1086.870 629.744 1174.785 559.241 2075.187

Page 22: Off-farm Income and Smallholder Commercialization: Evidence from Ethiopian Rural Household Panel Data By Tesfaye B. Woldeyohanes

22

Preliminary ResultsTable 3: Summary descriptive statistics - HH and HH head characteristics by year (n=1184) General

Background

Data and Emperical Models

Objectves

Hypothesis

Results

Problem statement

Variables 1997 1999 2004mean SD mean SD Mean SD

Age of household head 46.317 15.260 47.747 15.045 51.784 14.781Male household head 0.779 0.415 0.736 0.441 0.716 0.451Literate household head

0.258 0.438 0.259 0.438 0.280 0.449Family size (no) 6.053 2.744 5.332 2.563 5.748 2.526Available family labor 2.904 1.693 2.703 1.412 3.188 1.437Farm land owned(ha) 1.532 1.408 1.316 1.370 1.583 1.302Equine owned(no) 0.822 1.544 0.781 1.251 0.847 1.588Livestock owned(TLU) 3.279 3.635 2.883 2.818 2.992 3.218Distance to the nearest market (km) 12.764 5.877 10.684 4.853 8.541 5.858Involvement in extension program 0.061 0.239 0.115 0.319 0.147 0.354

Page 23: Off-farm Income and Smallholder Commercialization: Evidence from Ethiopian Rural Household Panel Data By Tesfaye B. Woldeyohanes

23

Thank you for your Attention!