The high poverty incidence in the county is a concern that needs to be addressed by our policy makers. Official poverty statistics from the National Statistical Coordination Board (NSCB) shows that the reduction in poverty over the past two decades has been quite dismal from 38% in 1988 to 26% in 2009 or less than one percent reduction per year. Since poverty incidence has dynamic patterns, studies using official poverty data encounter difficulty because of limited number of data points. This study builds econometric models in analyzing the movement of poverty in the country using the quarterly self-rated poverty series of the Social Weather Stations. The first model uses Markov Switching to determine the states of poverty. It assumes two states: high and moderate states of poverty. A high 61% of the population considered themselves as poor when the country is in the state of high poverty. In times of moderate poverty, 49.5% of the population considered themselves as poor. The result shows that once the country is in the state of high poverty, it stays there for an average of 24 quarters, or six years, before moving out. The paper then builds a logistic regression model to show what determines the states of high poverty. The model shows that a one-percent increase in agricultural output in the previous quarter reduces the probability of being in the high state of poverty by about 8 percentage points, all things being the same. The study shows that poverty incidence in the country is dynamic and frequent monitoring through self-rated poverty surveys is important in order to assess the effectiveness of the government programs in reducing poverty. The self-rated poverty surveys can complement the official statistics on poverty incidence.