Induced Wage Effects of Changes in Food Prices in Egypt

Abstract

The trend in real agricultural wages in Egypt over the two decades since the mid-1970s is well described by an inverted U-shaped curve with a peak around 1985. But the rise and fall of real wages masks a complex dynamic process by which nominal wages adjust in response to changes in food prices. We use governorate-level panel data for 1976–93 to explore the nature of this adjustment process. Our results indicate that nominal wages adjust slowly. There is a significant negative initial impact of rising food prices on real wages, though wages do catch up in the long run.

Notes

Gaurav Datt, Senior Economist, South East Asia and Pacific Region, World Bank Group, Level 19, 14 Martin Place, Sydney, Australia NSW 2000. E-mail: [email protected]. Jennifer C. Olmsted, Associate Professor, Department of Economics, Sonoma State University, 1801 E. Cotati Avenue, Rohnert Park, CA, 94928-3609, USA. E-mail: [email protected]. The research for this paper was undertaken as part of the Egypt Food Security Project supported by USAID Grant Number 263–G–00–96–00030–00, while both authors worked for the International Food Policy Research Institute. For useful comments or other forms of help, we would also like to thank Akhter Ahmed, Harold Alderman, Ragui Assaad, Heba El-Laithy, Lehman Fletcher, Nabil Habashi, Lawrence Haddad, Dean Jolliffe, Mohamed Omran, Alan Richards, Pieter Serneels, Anand Swamy, Yisehac Yohannes, participants at the IFPRI Workshop on Food Security Research Project in Egypt (Cairo, August 1997) and the Middle East Studies Association Meeting (San Francisco, November 1997), and this Journal's referees. We are particularly grateful to Jyotsna Jalan for valuable suggestions on the econometric work. The paper represents the views of the authors which should not be attributed to IFPRI or to the institutions to which the authors are currently affiliated.

Despite the general dearth of empirical work on this topic, a useful analysis for rural Bangladesh can be found in Ravallion [ 1987 ] and Boyce and Ravallion [ 1991 ].

For time–series evidence on the importance of agricultural wages as a determinant of rural poverty in India, see Datt and Ravallion [ 1998 ].

For a critical review of alternative models of wage determination for rural labour markets in developing countries, see Datt [ 1996 ].

Rural wage labour households have one of the highest poverty rates across all socio-economic groups in Egypt [ Datt, Jolliffe and Sharma, 1998 ].

This goes back to some of the early work by Hansen [ 1966 , 1969 ] who, for instance, questioned the usefulness of the subsistence wage theory in explaining the determination of Egyptian agricultural wages.

Instances of such unrest have occurred in Egypt before, most notably during the food riots of 1977 [Richards and Waterbury, 1996 , Gutner 1999 ].

See Datt and Olmsted [ 1998 ] for further details on the collation of this data set.

It should be noted that the investigation of the wage–price relationship was not the primary focus of de Janvry and Subbarao's study. A discussion of this study is nevertheless useful for motivating the key features of our approach.

The Ministry of Agriculture and Land Reclamation has been the key source of agricultural wage data for Egypt. Data from this source have been used in most studies on agricultural wages in Egypt, including Fitch, Ali and Mostafa [ 1980 ], de Janvry and Subbarao [ 1983 ], Assaad and Commander [ 1994 ], and Richards [ 1994 ].

The implications of measurement error in the wage data for our results are further discussed in Section A.2 in the Appendix.

However, there is some weak evidence of declining seasonality over time, which has also been sometimes noted in the literature; [see Richards 1994 , for instance]. In a model incorporating both seasonal effects and season–time interactions, we found that at the start of our period, wages were significantly higher (by 3 per cent on average) during the April–July and August–November periods (a significant positive common effect for these seasons); this is consistent with the known pattern of seasonality in the demand for agricultural labour. Season 1 corresponds fairly broadly to the lean season, which generally lasts from December to February [ Commander and Hadhoud, 1986 ]. But we also found the season–time interactions to be negative, though not significant. Upon eliminating the season–time interactions, the seasonal effects became altogether insignificant.

See Assaad and Commander [ 1994 ] and Richards [ 1994 ], for instance.

In the unorganised labour market setting, there is no indexation of wages and contracts have to be typically renegotiated to effect wage changes.

For different characterisations of the agricultural labour market in the Egyptian case, see de Janvry and Subbarao [ 1983 ], Grabowski and Sivan [ 1986 ], Commander [ 1987 ], and Richards [ 1994 ].

See Hendry [ 1995 ] on the generality of even a simple dynamic specification such as AD(1,1), which nests a variety of empirical dynamic models as special cases.

A detailed discussion of the data sources and the construction of the model variables can be found in Datt and Olmsted [ 1998 , Appendix 1].

The non-food component of the rural CPI was derived from the food and general indices using the average rural food consumption share 64.314 per cent obtained from the Income and Expenditure Survey 1990–91(calculated from the weighting diagram reported in CAPMAS [1996]).

There has been some concern about measurement error in CAPMAS’ CPI data, as for instance discussed by Cardiff [1997]. Cardiff's main argument is that CAPMAS should re-weight the CPI with a more recent consumption structure such as the one based on the 1995–96 Household Income, Expenditure and Consumption Survey. While there exists a case for such re-weighting for better monitoring of future changes in living standards and poverty, it is less relevant to our study which ends in 1993. For us what matters most is consistent measurement of changes in (rather than levels of) food and non-food prices, and even with regards to these, our estimation framework aims to control for certain types of measurement error (see Appendix for detailed discussion).

CAPMAS [1995]. For further details on the construction of this yield index and the data sources used, see Datt and Olmsted [ 1998 , Appendix 1]. The underlying area, production and price data have also been used by Rady et al. [ 1996 ] who also provide more details on these data.

Due to lack of data, we are unable to include variables representing labour demand originating in the informal sectors, such as services and construction. But, to the extent that economic activity in these sectors co-varies with that in the industrial sector, the public and private industrial output variables serve as partial proxies. There is also some indication of declining importance of construction as a competing source of labour demand: ‘… although construction work in the village or locality is often seen as a substitute for hired agricultural labour, the scope of work now open to rural households has considerably broadened. Indeed, a very small proportion of the sample was found to be working in the construction sector’ [ Commander, 1987 ].

See for instance, de Janvry and Subbarao [ 1983 ], Commander and Hadhoud [ 1986 ], Commander [ 1987 ], Adams [ 1991 ], Fergany [ 1991 ], Richards [ 1994 ], and Serageldin and Wouters [ 1996 ].

For convenience, we continue to use notation where all model variables are subscripted with both j and t, but for these variables x_jt = x_t .

We initially began with an AD(3,3) formulation, which seemed a natural choice given that our data set has three observations per year, but residual autocorrelation led us to introduce an additional lag.

See Arellano and Bond [ 1991 ] and Sevestre and Trognon [ 1996 ] for further discussion of this estimator for dynamic panel data models. The GMM estimator has been set up as a GAUSS-based programme; further details on the implementation of the estimator are given in Arellano and Bond [ 1988 ].

Thus, the variables x_jt are not only allowed to be correlated with the unobserved governorate-specific effects η_j, but they are also allowed to be contemporaneously correlated with the random error v_jt .

This test is based on the covariance between IV residuals and a set of instruments that need not have been used in the estimation. This covariance should be zero if the model is correctly specified, and the choice of instruments is valid.

The 20 deleted parameters related to the following variables: lagged nominal wages, Δw_− 2, Δw_− 4 ; lagged food price index Δp^f _− 3 ; current and lagged yield per feddan, ΔYLD_− 1, ΔYLD_− 3 ; current and lagged total cropped area, ΔAREA, ΔAREA_− 1, ΔAREA_− 3 ; lagged total population, ΔPOP_− 3 ; current and lagged public sector industrial output per person, ΔYPUB, ΔYPUB_− 3 ; lagged private sector industrial output per person, ΔYPVT_− 1, ΔYPVT_− 2, ΔYPVT_− 3 ; lagged real exchange rate, ΔXR, ΔXR_− 1 ; lagged real remittances, ΔREMIT_− 1, ΔREMIT_− 2 ; and seasonal dummy variables, ΔSEAS2, ΔSEAS3. See Datt and Olmsted [ 1998 ] for details.

The observed lack of seasonality relates to real wages insofar as the model is estimated conditional on prices. There is greater evidence of seasonality in nominal wages which is masked by a similar seasonality in prices. For instance, for Bahera governorate, nominal wages during April–July and August–November were on average about 11 per cent higher than during the relatively lean months of December–March. This gets muted (to 6 per cent) for real wages because there is also a similar seasonality in prices. Seasonal effects are of course defined in terms of the four-month intervals introduced earlier (Section 3) which may not strictly reflect the agricultural peak and slack seasons across governorates.

This data issue is discussed further in Datt and Olmsted [ 1998 , Appendix 1].

Thus, for instance, for a model with w_t − 3 as a regressor, valid instruments could be based on w_t − 5 .

Similar arguments also apply to potential measurement error related to the possible decline in the in-kind component of wages that has been noted by some observers.

Additional information

Notes on contributors

Gaurav Datt

Gaurav Datt, Senior Economist, South East Asia and Pacific Region, World Bank Group, Level 19, 14 Martin Place, Sydney, Australia NSW 2000. E-mail: [email protected]. Jennifer C. Olmsted, Associate Professor, Department of Economics, Sonoma State University, 1801 E. Cotati Avenue, Rohnert Park, CA, 94928-3609, USA. E-mail: [email protected]. The research for this paper was undertaken as part of the Egypt Food Security Project supported by USAID Grant Number 263–G–00–96–00030–00, while both authors worked for the International Food Policy Research Institute. For useful comments or other forms of help, we would also like to thank Akhter Ahmed, Harold Alderman, Ragui Assaad, Heba El-Laithy, Lehman Fletcher, Nabil Habashi, Lawrence Haddad, Dean Jolliffe, Mohamed Omran, Alan Richards, Pieter Serneels, Anand Swamy, Yisehac Yohannes, participants at the IFPRI Workshop on Food Security Research Project in Egypt (Cairo, August 1997) and the Middle East Studies Association Meeting (San Francisco, November 1997), and this Journal's referees. We are particularly grateful to Jyotsna Jalan for valuable suggestions on the econometric work. The paper represents the views of the authors which should not be attributed to IFPRI or to the institutions to which the authors are currently affiliated.