Ethiopian Development Research Institute and International Food Policy Research Institute (IFPRI/EDRI), Tenth International Conference on Ethiopian Economy, July 19-21, 2012. EEA Conference Hall
1. ETHIOPIAN DEVELOPMENT
RESEARCH INSTITUTE
URBAN WAGE BEHAVIOR DURING
FOOD PRICE HIKES:
THE CASE OF ETHIOPIA
Derek Headey, Fantu Bachewe, Ibrahim Worku,
Mekdim Dereje & Alemayehu Seyoum Taffesse
IFPRI ESSP II
Ethiopian Economic Association Conference
July 19-21, 2012
Addis Ababa
1
2. Background
• The global food crises of 2007-08 and 2010-11
sparked a number of efforts to understand the
poverty impacts of higher real food prices
• On the one hand World Bank simulation approaches
suggested global poverty rose by 160 million people
• However, subjective survey data from Gallup suggest
substantial variation of impacts, and that strong
economic growth in developing countries limited the
impacts of higher prices (Headey 2011)
• A third less common approach is to deflate wages by
(food) prices as a proxy for disposable income
3. Background
• Some precedent on agricultural wages & food prices
• Literature is almost solely confined to Bangladesh
(Ravallion; Palmer-Jones; Rashid), and Philippines (Lasco et al.).
• Bangladesh: limited short run impacts of prices on
wages; Philippines: fairly large short run impacts
• More recent study by Mason et al. (2010) looks at
urban manufacturing wages in Zambia and Kenya.
No econometrics, but “food-disposable” wages fell in
2008, but were still high by historical terms because
of strong economic growth.
4. Background
In this paper we have two objectives:
1. To track real wages in (as per Mason et al.)
2. To formally test wage adjustment (as per Lasco et al., etc)
Our context – Ethiopia - is a particularly interesting one:
1. Very poor (60% of urban pop. with <$2/day; 20% uN rate)
2. Very understudied in World Bank & Gallup studies
3. Unusually, we have monthly panel data on informal
or casual wages (much better than previous data)
4. Arguably the most rapid food inflation in the world
in 2008 and 2011
5. Figure 1. Average monthly inflation in Ethiopia relative
to other developing countries: 2004-2011
4.0 3.0
Food inflation (%) Relative food inflation (%)
2.0
3.0
1.0
2.0
0.0
1.0
-1.0
0.0 -2.0
2005
2004
2006
2007
2008
2009
2010
2011
2011
2004
2005
2006
2007
2008
2009
2010
-1.0
Other developing countries Ethiopia
Source: ILO (2012).
6. 2. Data and methods
• CSA consumer price data from 115 “urban” markets
around the country, from July 2001 to October 2011
• In addition to prices on food & non-food items, CSA
asks about daily laborer wages, maids wages, guards
• But since maids and guards are partly paid with food-
in-kind, we only focus on laborers (trends the same)
• Prices and wages collected for 3 respondents (firms
or households) in each market and then averaged
• Enumerators try to measure the same respondents
(kind of a panel?)
7. 2. Data and methods
• To create a better wage welfare proxy, we create food
and non-food price indices specifically for the poor
• We used the 2004/05 HICES expenditure data, and
measure expenditure shares just for the bottom 40%
• We do this for rural and urban areas of each region,
then apply these weights to the CSA price data to
derive a set of spatially disaggregated “poor person’s
price indices” (PPPIs) for food, non-food and all items
• We deflate laborer’s wages by both food prices and
total prices for the poor.
8. 2. Data and methods
• In principle, deflating by total prices is most
appropriate for welfare interpretation, but deflating
food prices may be more relevant for ultra-poor who
may spend almost all of their income on food
• More generally, are daily laborer’s wages are a good
welfare indicator for the poor?
• For India, Deaton and Dreze argue that wage series
for casual labor are a good poverty indicator, because
they represent the reservation wage for the poor
• For urban Ethiopia we make the same argument
9. 2. Data and methods
• Finally, we use panel regressors to see whether wages
react to food prices in the short run
• We use a panel vector error correction model (PVEC)
& spatially disaggregate by town/city size & regions
• PVEC effectively separates out a long run adjustment
relationship (cointegrating relationships) and short
run adjustments.
• We are more interested in the short run adjustments
as they are more welfare-relevant. “In the long run we
are all dead”
12. Figure 3. Trends in real daily laborer wages deflated by the
urban poor’s food and total prices indices
13
Wages deflated by poor person's food CPI
21%
Real daily wage of laborers (Dec. 2006 birr)
Wages deflated by poor person's total CPI fall
12
10%
fall
11
10
9
26% fall 26% fall
8
7
Source: Author’s calculations from CSA (2011b) data. See section 2 for methods used.
15. • The long run relationship shows substantial
adjustment of wages to food prices (elasticity of
greater than 1), but not to non-food prices:
• Wages = -2.9 +1.2*Food CPI -0.1* Nonfood CPI -0.001* t
However, it is obviously difficult to put a welfare
interpretation on this equation
• Especially since the short run results show scarcely
any adjustment . . .
17. 4. Conclusions
Main findings:
• Casual workers in urban Ethiopia have been hit hard by
rapid food inflation in 2008 & 2011, particularly ultra-
poor:
10-26% loss of disposable income year, region, indicator
• 2011 crisis (ongoing) seems worse than 2008 crisis
• Given that households could have many coping
mechanisms (e.g. longer working hours), these may be
upperbound estimates of welfare impacts
18. 4. Conclusions
Policy questions:
• Govt. has focused on trying to directly curb food
inflation through price controls & subsidization
• Efforts to reduce domestic inflation are sensible, but
the capacity to fully reduce inflation may be limited
given higher international prices and ambitious
domestic growth scenarios in the GTP
• So does Ethiopia need an urban social safety net?
• Many considerations here, but one option is to index
cash transfers to our poor person’s price index
19. 4. Conclusions
Research implications
• Further work could try to validate the wage series as
a relevant and accurate welfare indicator for poor
• CSA could consider asking about food-in-kind for
maids salaries, guards salaries
• Arguably the collection of wage series by statistical
agencies elsewhere should be scaled up
• They appear to be a cost-effective and very useful
high frequency indicator of urban welfare, and in
some contexts, agricultural welfare (e.g. South Asia)
Notes de l'éditeur
Can’t promise I am going to rock your world, but I hope to promote some debate and some further research or feasibility analysis into whether these proposals are really viable.
First panel: Excludingzimbabwe, Ethiopia had the highest food inflation in the world in 2008. Second paenl – this is the change in food CPI/nonfood CPI.
Deaton and Dreze (2002) argue that for India and South Asia agricultural wages are also a good poverty indicator for the rural poor because in South Asia agricultural labor is the reservation occupation for the poor