本研究欲探討在同期相關之下失業對糧食不安全的影響，資料取自美國普查局、美國勞動統計局、美國經濟分析局、美國農業部及聯邦房屋金融局，本研究主要使用2005年至2015年美國五十州及華盛頓特區失業率、貧窮率及實質人均所得對糧食不安全進行探討分析。糧食不安全分為三個層次，本研究分別為近乎糧食不安全、糧食不安全及低度糧食不安全來進行估計，有別於文獻，本研究探討共同衝擊下，失業率對糧食不安全三層次的影響，採用Pesaran之一般性相關效果混和估計式(CCEP)及一般性相關效果平均估計式(CCEMG)以整體平均的概念來消除這些未觀察到共同因子之同期相關。失業率會對糧食不安全造成極大的影響，然而在去除掉同期相關的情況下，我們發現失業率僅對近乎糧食不安的家戶有影響。 同時本研究也控制人均食品及飲品實質支出、家用食品物價指數、房價指數及營養補充援助計畫作為敏感性分析，本研究以工具變數控制營養補充援助計畫對糧食不安全三層次的內生性。結果顯示：加入敏感性分析變數後，能夠提升失業率對糧食不安全三層次之顯著程度，使失業率不論在任何群集皆為正向且顯著。然而使用CCEP及CCEMG後會把顯著效果消除，因此我們會將重點解釋變數進行排列組合個別對糧食不安全三層次進行研析，並發現CCEP估計式衡量的效果會比CCEMG估計式結果好，而這是因為CCEP估計式較適合我們的樣本大小，控制人均食品及飲品實質支出會比實質人均所得更能顯示出失業率對糧食不安全的影響。 This study is intended to investigate the impact of unemployment on food insecurity in common shocks, and the data are provided by the US Census Bureau, the US Bureau of Labor Statistics, the US Bureau of Economic Analysis, US Department of Agriculture and the Federal Housing and Finance Agency. We mainly use the data of unemployment rate, poverty rate and per capita real GDP from 2005 to 2015 in US states and Washington DC to analyze the impact to food insecurity. Food insecurity in three levels are Marginally food insecurity, Food insecurity and Very low food security, respectively. The contribution of our study is that we analyze the effect of unemployment rate to three levels of food insecurity under common shocks. Then we find that unemployment rate has great impact to food insecurity. After we used common correlated effected estimator to eliminate Cross-Section Dependent, we also find that unemployment rate is only statistically significant for marginally food insecurity households. After all, we use per capita real food and beverage expenditure by State, food at home consumer price index, house price index and SNAP subsidy rate as controlling variables in the sensitivity analysis. We find that that adding those controlling variables of sensitivity analysis strengthen the impact of unemployment rate to all three levels of food insecurity; that is, the estimates become more significant. After then we estimate the various models by controlling different sets of variables using CCEP and CCEMG estimators. We find that CCEP estimator is more appropriate than CCEMG estimator to fit our model. And adding per capita real food and beverage expenditure is more capable of strengthening the impact of unemployment rate to three levels of food insecurity statistically than per capita real GDP.
