# Multistate Markov model for household’s food insecurity transitions and their influencing predictors during COVID-19 pandemic in Ethiopia

## TL;DR

This study uses a Markov model to analyze how households in Ethiopia transitioned between food insecurity states during the pandemic and what factors influenced these changes.

## Contribution

The study introduces a multistate Markov model to analyze food insecurity transitions and their predictors during the pandemic in Ethiopia.

## Key findings

- Households initially in severe food insecurity had a 94.8% chance of remaining food insecure.
- Female-headed households were more likely to transition from food secure to mild and from mild to moderate food insecurity.
- Income loss and assistance provisions significantly predicted food insecurity for households in food secure and mild states.

## Abstract

Food insecurity is one of the most widespread social problems that millions of the world’s people are experiencing worries. The problem has been rising, particularly in Ethiopia, since 2020 due to Covid-19 with changing households’ food insecurity dynamics which is not clear yet. This study aimed to investigate the dynamics of households’ food insecurity transitions and assess the associated covariates during Covid-19 pandemic in Ethiopia. The study used Ethiopia-High Frequency Phone Survey longitudinal data set collected by World Bank with the total of 13517 observations repeated over time. Multistate Markov model was employed to model the paths of changing household’s food insecurity process and identify the influencing covariates of each transition. The result indicated that out of households observed in food secure state initially at one visit, 26.0% were observed in food insecure states. The corresponding values for households that were initially at mild, moderate, and severe states were 68.6%, 88.3%, and 94.8% respectively. Once a household enters mild and moderate, his incidence rate increases revealing that a household initially at severe food insecure states experiences higher risk of staying as food insecure. Sex of household head is significantly associated with food insecurity; with female headed households are respectively 1.2345 and 1.1720 times more likely to enter from food secure to mild and from mild to moderate states as compared to male headed households. Households’ residence, income loss, employment status, and assistance provisions during Covid-19 were significant predictors of food insecurity given that they were primarily at food secure and mild food insecure states; but they lack the significance for households initially at moderate state. The initial food insecurity state of households, time duration, and varying effects of covariates are the crucial issues in predicting food insecurity dynamics. Timely interventions are necessary for the effective recovery of household’s food insecurity.