1. Dennis S. Mapa Kristelle M. Castillo Krizia DR Francisco School of Statistics University of the Philippines, Diliman 52 nd Annual Meeting Philippine Economic Society (PES) 14 November 2014 Hotel Intercontinental, Makati CIty

2.  This paper looks at the factors that influence the dynamic nature of hunger incidence in the Philippines using the self-rated data from the SWS quarterly surveys on hunger.  Variables identified as potential determinants of hunger incidence are, among others, changes in the price of rice and job misery index (sum of the employment and unemployment rates).  In this paper, the authors used Vector Auto Regressive (VAR) model and Time-Varying Parameter (TVP) model. Objectives

3. Quarter Year 20072008200920102011201220132014 1st3.46.912.34.24.62.73.55.7 2nd2.814.35.53.55.42.43.16.5 3rd2.719.30.23.75.23.1 3.5 6.8 4th4.115.13.43.15.23.0 4.8 Inflation Rate of the Poorest 30 Percent of Households Weight of Food in the Basket: 70 percent (vs. 47 percent for all households); Weight of Rice: 23 percent (vs. 9 percent for all households)

7. Vector AutoRegressive (VAR) Model We can generalized the mathematical representation of a VAR model as: where y t is a (k x 1) vector of endogenous variables, A 1, A 2,…, A p are matrices of coefficients to be estimated, and e t is a (k x 1) vector of innovations that may be contemporaneously correlated but are uncorrelated with their own lagged values and uncorrelated with all of the right-hand side variables. e t is assumed to be normally distributed with mean 0 and covariance matrix Σ.

8. VariableTest StatisticP-valueRemarks HWEGF (in nat. log.)-2.8531440.1856I(1) MISERY (in nat. log.)-6.5450780.0000Stationary* MISERY_SA (in nat. log.)-3.4566380.0546Stationary* RICE (in nat. log.)-2.5608980.2992I(1) TOTAL (in nat. log.)-2.8117370.1998I(1) *I(0) + trend: Stationary with deterministic trend. Test for Unit Root (Augmented Dickey-Fuller)

9. DLOG(TOTAL)DLOG(RICE)DLOG(HWEGF)LOG(MISERY_SA) DLOG(TOTAL(-1))-0.432*-0.003 0.010 (0.123)(0.016)(0.005)(0.041) [-3.505][-0.202][-0.565][ 0.239] DLOG(RICE(-1))2.043*0.180-0.0200.018 (1.065)(0.139)(0.0430)(0.358) [ 1.919][ 1.301][-0.463][ 0.051] DLOG(HWEGF(-1))-5.6610.7110.297*0.637 (3.507)(0.456)(0.142)(1.179) [-1.614][ 1.558][ 2.096][ 0.541] LOG(MISERY_SA(-1))0.685*-0.0680.0090.381* (0.375)(0.049)(0.015)(0.126) [ 1.825][-1.401][ 0.602][ 3.017] C-2.2280.230-0.0232.054 (1.247)(0.162)(0.050)(0.419) [-1.786][ 1.415][-0.448][ 4.898] R-squared0.30680.11720.10240.1831 Adj. R-squared0.24780.04210.02600.1136 Standard Errors are in ( ) and t-statistics in [ ] *significant at critical value = 1.75in t-statistics

10. Response of Total Hunger Incidence to One Standard Deviation Increase in Rice at Quarter t

11. Response of Total Hunger Incidence to One Standard Deviation Increase in Job Misery at Quarter t

12. Time-Varying Parameters (TVP) Model A linear (Gaussian) state space representation of the dynamics of the (n x 1) vector y t is given by the system of equations: y t = X t β t + ε t,ε t ~N(0, Σ ε ) measurement equation(1) β t = c t + Fβ t-1 + ν t ν t ~N(0, Σ ν ) transition equation(2)

13. The TVP model using state space representation: ΔLOGTOTAL t = β 0,t + β 1,t ΔLOGRICE t-1 + β 4,t LOGMISERY_SA t-1 + β4,t ΔLOGHWEGF t-1 + β 5,t ΔLOGTOTAL t-1 + ε t,ε t ~ i.i.d. N(0, σ 2 ) whereβ i,t = β i,t-1 + ν it,ν it ~ i.i.d. N(0, σ 2 i )i = 0, 1, 2, 3, 4, 5 There is one signal equation and five state equations. Assume σ 2 i = 1 for all i = 0, 1, 2, 3, 4, 5 are the fixed parameters while β i,t are the time-varying parameters. Time-Varying Parameters (TVP) Model

14. Movement of the Coefficient of Rice in the TVP Model average from 2002Q1 to 2008Q4: 1.02 vs average from 2009Q1 onwards: 1. 88

15.  Results from the VAR model show that a shock (increase) in the price of rice at the current quarter tends to increase hunger incidence in the succeeding quarter.  A shock (increase) in job misery index at the current quarter also increases the hunger incidence in the next quarter.  Analysis using the time-varying parameter (TVP) model shows a higher effect of changes in the price of rice to hunger incidence after the global rice crisis in 2008.  This means that hunger incidence is becoming very sensitive to changes in the price of rice. Summary of Results