Abstract

This paper has examined the long-run relationship between daily per capita calorie intake, per capita income and food prices for Pakistan using aggregate data 1960-2001. Cointegration analysis yields the income-calorie elasticity of 0.21, while food-price elasticity is insignificant. Thus, economic growth, as measured by increasing per capita income, has significantly improved calorie intake in Pakistan; future economic growth can alleviate further inadequate calorie intake. However, significant improvements in calorie intake cannot be made directly by food subsidies. Nevertheless, the policies that lower food prices also have the indirect effect of increasing real incomes via the income effect. So it is in this sense that foodsubsidy policies may have a role in improving calorie intakes. In the context of access to food, it would be important to identify the food insecure people, which are financially poor and are unable to acquire sufficient food, even if the overall supply of food in the country is sufficient. Further, causality tests indicate a bidirectional relationship from income to calorie intake; and from calorie intake to income. Key words: Per capita calorie, cointegration, Granger-causality, Pakistan Introduction In Pakistan per capita calories intake has grown from 1940 (kilo) calories per day in 1960 to 2457 in 2001 with an average annual growth rate of 0.60 per cent (Govt. of Pakistan, 2003). Notwithstanding this increase, food security remains an unfulfilled dream for currently about 42 million people (UN, 2001). The fact that about one third of the population does not have access to food needed for adequate nutrition is manifested by the incidence of malnutrition. Among the 174 nations covered by the latest survey using the criteria of Human Development Index (HDI), Pakistan was ranked as number 135 (UN, 2001). Access to food is mainly related to per capita income, and for the last four decades Pakistan has been trying to increase the per capita income. Food poverty (calorie based) incidence showed that about one-third of the households are living below the food poverty line (consuming calories below the recommended level) and they are not meeting their nutritional requirements. The incidence of food poverty is higher in rural areas (35%), than in urban areas (26%) (UN, 2001). The relationship between calorie intake and income therefore is crucial and much of the literature on Many time series are non-stationary and in general OLS malnutrition has focused on this relationship but regressions between non-stationary data are spurious. estimated calorie-income elasticity vary considerably The presence of unit roots in the autoregressive (Bouis, 1994, summarizes this literature). There are two representation of a time series leads to non-stationarily, lines of inquiry: first, does calorie intake rise with and such series, referred to as being integrated of order income, and second, is income generation affected by one [I (1)], must be first-difference to render them calorie intake? The former focuses on the estimation of stationary (or integrated of order zero). Where [I (1)] calorie-demand relationship, while the latter is at the series move together and their linear combination is centre of the efficiency wage hypothesis. It is therefore stationary, the series are cointegrated and the problem clear that causality in the calorie-income relationship of spurious regression does not arise. Cointegration can run in either or both directions. It is evident from the literature on the efficiency wage hypothesis where there is concern about the endogeneity of income (see for example, Strauss 1986; Sahn and Alderman, 1988; Haddad and Bouis, 1991 and Behrman et al., 1997). Recently, Dawson and Tiffin (1998); Dawson (2002) examined the long-run calorie-income relationship for India and Pakistan and the estimated elasticities are 0.34 and 0.19 respectively. However, notwithstanding the fact that food subsidies are common in LDCs and the estimates with respect to food prices in general, tend to be significant. Our aim here is to re-examine the long-run relationship between per capita calorie intake, per capita income, and food prices using aggregate time series data and cointegration analysis for Pakistan and to test for the direction of causality between calories and income. The remainder of the paper is organized as follows: Section 2 discusses our empirical methodology, Section 3 discusses the data and results, and Section 4 summarizes and concludes. Materials and Methods Mushtaq et al.: An Examination of Calorie Demand Relationship in Pakistan 160 implies the existence of a meaningful long-run while n is the optimal lag length orders of the variables equilibrium (Granger, 1998). Since a cointegrating which are determined by using the general-to-specific relationship cannot exist between two variables which modeling procedure (Hendry and Ericson, 1991). Our are integrated of a different order, we first test for the null hypotheses are as follows. Y will Granger cause C order of integration of the variables. if $ …0. Similarly, C will Granger cause Y if F 0. There In testing for the presence of unit roots in the individual will be bidirectional causality if $ …0 and F 0. To time series using the augmented Dickey-Fuller (ADF) implement the Grnger causality test, F-statistics are test (Dickey and Fuller, 1981; Said and Dickey, 1984), calculated under the null hypothesis that in Equ (2) and both with and without a deterministic trend, we follow the (3) all the coefficients of $ , F = 0. sequential procedure of Dickey and Pantula (1987): the null of the largest plausible number of unit roots, assumed to be three, is tested and, if rejected, that of two unit roots is tested and so on until the null is not rejected. The number of lags in the ADF-equation is chosen to ensure that serial correlation is absent using the Breusch-Godfrey statistic (Greene, 2000, p.541). If they are integrated of the same order, Johansen’s (1988) procedure can then be used to test for the presence of a cointegrating vector between calorie intake, income and food prices. The procedure is based on maximum likelihood estimation of the error correction model: aZ = *+' aZ ' aZ + ..... ' aZ +Bz + μ (1) t 1 t-1 + 2 t-2 p-1 t-p+1 t-p t Where Z = [C , Y, P ], C is per capita calorie intake, Y is t t t t t t income, and P is the price of food, = z z and B and t t t t-1 ' are (nxn) matrices of parameters with ' = (I -A -A --i i 1 2 ...A), (I = 1,... k 1), and B = I B B --B . The term Bz i 1 2 k t-p provides information about the long-run equilibrium relationship between the variables in z . Information t about the number of cointegrating relationships among the variables in z is given by the rank of the B-matrix: if B t is of reduced rank, the model is subject to a unit root; and if 0<r<n, where r is the rank of B, B can be decomposed into two (n x r) matrices and $, such that B = $' where $'z is stationary. Here, is the error t correction term and measures the speed of adjustment in )zt and $ contains r distinct cointegrating vectors, that is the cointegrating relationships between the nonstationary variables. The Johansen procedure estimates (1), and trace statistics are used to test the null hypothesis of at most r cointegrating vectors against the alternative that it is greater than r. If cointegration is established, then Engle and Granger (1987) error correction specification can be used to test for Granger causality. If the series Ct and Yt are both I (1) and cointegrated, then the ECM model is represented by the following equations. Where ) is the difference operator μ and , are the white t t noise error terms, ECT is the error correction term t-I derived from the long-run cointegrating relationship., j i

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call