Did The Commodity Price Spike Increase Rural Poverty: Evidence From Bangladesh

Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 32

Did the commodity price spike increase rural poverty: Evidence from Bangladesh

Joseph V. Balagtas, Purdue University Humnath Bhandari, IRRI Sam Mohanty, IRRI Ellanie Cabrera, IRRI Mahabub Hossain, BRAC

200

300

400

500

600

700

800

100

Jan-00 Aug-00 Mar-01 Oct-01 May-02 Dec-02 Jul-03 Feb-04 Sep-04 Apr-05 Nov-05 Jun-06 Jan-07 Aug-07 Mar-08 Oct-08 May-09 Dec-09 Jul-10 Feb-11

Background: Rice Prices, 2000-2011 (2000 US$)

Thai Export Bangladesh Wholesale

Effects of the Global Rice Turmoil (and other events) on Rural Poverty
Rice market turmoil has potentially mixed effects on rural poor:
higher price of staple food harms consumers higher price of crop potentially boosts income for farm households
Directly for those farms with marketable surplus Indirectly for those farms consuming own production

Net effect is an empirical question

Research Question and Approach


How did the market events of 2007-08 affect rural households in Asia?
Focus on household income and poverty

Long-term household survey of rural households in Bangladesh allows crucial insights

Approach: Evidence from Panel of HH in Bangladesh


We update and extend work by Nargis & Hossain (2006) who analyze 1988-2004 data
Find incidence and depth of poverty declined over that time Find that diversification away from agriculture, overseas migration strongly associate with higher income

We use a new round (2008) of the same survey to:


document further changes in poverty evaluate determinants of income and poverty

Data
Repeat survey of nationally representative sample of rural households in Bangladesh
Designed specifically to track rural livelihoods

Multistage random sampling


1 union randomly selected from each district (64) 1 village in each union selected based on land size and education Census of the village, followed by selection of households stratified on land ownership & tenure

64 Rural Villages in Bangladesh ~20 HH per village + their offshoots Four rounds collected: 1987, 2000, 2004, 2008

The Policy Setting


Last survey round (2008) occurs at the peak of the rice market turmoil Also other macroeconomic events
Financial crisis/global recession. Potential impact on income opportunities overseas, as well as returns to domestic business/services

Select Sample Statistics I


1988 2000 2004 2008

Sample size Farm size (ha) Nonfarm HH (%) Area under tenancy (%) Ag workers (no.) Domestic migrants (no.) Overseas migrants (no.)

1,238 0.62 34 22 1.2 0.2 0.01

1,872 0.54 42 33 0.9 0.4 0.1

1,927 0.48 39 40 1.0 0.4 0.1

2,010 0.47 43 14 0.8 0.5 0.2

Select Sample Statistics II


1988 2000 2004 2008

HH income (000 04 Taka) Per cap income (000 04 Taka) agriculture (%) rice (%) trade/business (%) remittances (%) Phys. Capital (000 04 Taka) agricultural (%)

65 11 58 26 9 5 17 47

78 14 43 16 21 13 22 37

82 16 44 15 19 14 36 48

95 19 43 15 15 23 44 32

The data reflect diversification away from agriculture:


Labor moving from farms to cities and abroad Balance of phys. capital tilting away from ag

How do these and other demographic shifts affect income?


Estimate an income regression to assess determinants of income

Determinants of log HH income, 2008


Independent variable Estimated coefficient

Owned land (ha) Non-farm HH No. of nonag workers Edu. of nonag workers Migrant workers, domestic Migrant workers, overseas Rice land in modern varieties N = 2010 R2 = 0.54

0.20*** -0.08* 0.16*** 0.03*** 0.03*** 0.50*** 0.06***

Determinants of HH Income
Results similar for all four survey waves Consistent with previous literature: income diversification raises income Corollary policy recommendation: invest in human and physical capital that allows HH to tap non-ag income opportunities (and diversify away from agriculture)
A pathway out of poverty

A potentially important caveat


CS studies potentially suffer from endogeneity
Unobserved HH household characteristics (e.g., ability, industriousness) potentially confound identification of causal effects

We estimate a fixed effects model on a subset of the sample that we observe in all four survey waves (~ 1000 households)
Uses dummy variables to control for timeinvariant, unobserved HH characteristics

FE Estimates of Income Determinants


Independent variable 2008 CS Fixed Effects

Owned land (ha) Non-farm HH No. of nonag workers Edu. of nonag workers Migrant workers, domestic Migrant workers, overseas Rice land in modern varieties

0.20*** -0.08* 0.16*** 0.03*** 0.03*** 0.50*** 0.06***


N = 2010 R2 = 0.54

0.18*** -0.12*** 0.194*** -0.004* 0.03** 0.29*** 0.03


N = 964, T = 4

Turning our attention to poverty


But first, any questions?

Measuring Poverty

Choice of poverty threshold, z


We calculate z based on FAO concept of moderate poverty
2100 kcal per day per person Data on rural food prices 30% of total income for non-food expenditures

This approach assumes no price response in consumption. Probably ok for staple goods and broad-based price increases. But worth noting.

Rural Poverty, Bangladesh, 1988-2004


1988 2000 2004

Per cap. income (000 04 Taka) Est. poverty line (000 04 Taka) Poverty head count (%) Poverty gap (Poverty gap)2
1. Incomes rising, poverty line steady 2. Poverty incidence decreasing 3. Depth of poverty decreasing

11 8.3 62 26.4 14.4

14 7.6 48 19.1 10.2

16 8.3 44 16.5 8.5

Rural Poverty, Bangladesh, 1988-2008


1988 2000 2004 2008

Per cap. income (000 04 Taka) Est. poverty line (000 04 Taka) Poverty head count (%) Poverty gap (Poverty gap)2

11 8.3 62 26.4 14.4

14 7.6 48 19.1 10.2

16 8.3 44 16.5 8.5

19 11.9 56 21.9 11.1

1. Incomes rising, but higher food prices drive poverty threshold up by 30% 2. Poverty incidence rises by 25% 3. Depth of poverty worsens

The Commodity Price Spike and Rural Poverty


12 percentage point increase in poverty incidence Approx. 13 million people fall into poverty in rural Bangladesh Suggests huge impact of macro events on rural poverty in Asia
We dont quite have the information to make the causal link.

How did the macroeconomic events affect poverty?


Do they close off pathways out of poverty, i.e., increase the incidence of chronic poverty Do they increase transitional poverty?

What HH characteristics or activities influence the outcome? Did diversification away from ag help?
To answer these questions we look at a subsample of our data, ~ 1000 HH that we observe in all four survey years.

The Poverty Mobility Matrix


Tracks poverty mobility between two years Four categories
poor that stay poor poor that move out of poverty nonpoor that fall into poverty nonpoor that remain nonpoor

Poverty Mobility Matrix, 2000-2004


2004 (row %) 2000 (col %) Poor (col %) Non-poor (col %) Total (col %) Poor (row %) 270 (68) (60) 127 (32) (25) 397 (100) (41) Non-poor (row %) 176 (31) (40) 391 (69) (75) 567 (100) (59) Total (row %) 446 (46) (100) 518 (54) (100) 964 (100) (100)

Of 446 poor HH in 2000, 60% remain poor in 2004; 40% escape poverty. Of 518 nonpoor HH in 2000, 25% backslide into poverty by 2004. Net reduction in poverty incidence (446 HH to 397 HH)

Poverty Mobility Matrix, 2004-2008


2008 (row %) 2004 (col %) Poor (col %) Non-poor (col %) Total (col %) Poor (row %) 255 (58) (64) 183 (42) (32) 438 (100) (45) Non-poor (row %) 142 (27) (36) 384 (73) (68) 526 (100) (55) Total (row %) 397 (41) (100) 567 (59) (100) 964 (100) (100)

Increase in chronic poverty from 60% to 64%. Among non-poor in 2004, 7 percentage point (28%) increase in transitory poverty. Net increase in poverty incidence.

Summarizing the Impact of the 2007-08 Events on Rural Poverty


Increased incidence of poverty
Less upward poverty mobility more transitional poverty.the commodity price spike and other macro events pulls more non-poor into poverty

Additional research questions


In the paper we show the chronic poor are more reliant on ag. Did the chronic poors reliance on agriculture reduce their ability to escape poverty? A probit model of poverty:
Pr[Poor(t)] = f(Poor(t-1), Ag(t-1), X(t-1))

Probit model of Poverty in 2004, Marginal Effects


Variable 2004

Poverty(t-1) Ag Income share(t-1) Land owned(t-1) Overseas migrant(t-1) Nonfarm HH(t-1) Tenant HH(t-1)

0.19*** 0.18** -0.10*** 0.03* -0.02*** 0.17***

Being poor in 2000 increases the chance youll be poor in 2004. Having more of your income from agriculture in 2000 also makes it more likely youll be poor in 2004. Consistent with previous work that suggests diversification away from ag Is a pathway out of poverty.

Probit model of Poverty in 2004 and 2008, Marginal Effects


Variable 2004 2008

Poverty(t-1) Ag Income share(t-1) Land owned(t-1) Domestic migrant(t-1) Overseas migrant(t-1) Nonfarm HH(t-1) Tenant HH(t-1)

0.19*** 0.18** -0.10*** 0.03* -0.02*** 0.17*** 0.10*

0.22*** -0.14** -0.14** -0.07*** -0.22*** 0.12** 0.19***

Migrant labor significantly shielded households from the 2007-08 events. But ag income also appears to have reduced poverty in 2008. Problem for renters.

Summary of Key Findings I


Rice price spike + global recession reversal of long trend out of poverty Increased poverty incidence
More chronic poverty and more transitory poverty

Chronic poor are more reliant on agricultural income, and increasingly so


Nonfarm households and land renters tend to be poor

Summary of Key Findings II


Appears that ag income and land shielded households from commodity price spike and global recession Either as income source, or as a hedge against high price of staple food This worked better for landowners; problem for renters Determinants of income/poverty are not time-invariant A cautionary tale for those recommending non-ag income sources as pathway out of poverty

Maraming salamat po! Questions?

You might also like