This dissertation consists of three essays that contribute to both applied and theoretical microeconomics. The first two essays provide a theoretical framework, empirical evidence, and an empirical strategy for a better understanding of the seasonality of food insecurity in developing countries, with a special focus on seasonal price changes of staple foods. More specifically, the first essay constructs a theoretical model to analyze how seasonal price changes of a staple food affect farmers' seasonal consumption in developing countries, where storage of the staple food can be used to smooth consumption. Crucially, sharp increases in the price of the staple food just before harvest can be viewed as a high return to savings, and this has important implications for interpreting the consumption and savings behavior of poor rural households. Then, the second essay addresses whether and how farmers smooth their consumption within a crop year, using three years of weekly household panel data from rural Zambia. Given seasonal price changes of the staple food, maize, some farmers buy it when prices are low and store it for consumption during the hunger season, while others run out of the staple food before the next harvest, and so buy it when prices are high. Results indicate that the former group successfully smooths its consumption, while the latter group reduces consumption during the hunger season in response to a negative harvest at the end of the previous crop year, and the effect of these negative harvest shocks produces an inverse U consumption pattern during the crop year, especially for farmers with few assets. These farmers reduce their consumption of non-staple foods and thus reduce their food diversity to maintain consumption of the staple food in the hunger season in spite of its price hike in that season. The third essay proposes an empirical strategy (the network approach) to analyze complex interactions among several agents, and illustrates how this approach works by applying it to the analysis of soccer games. By using a longitudinal data set of all soccer players in the top German league (the Bundesliga) over the course of ten seasons (2000/01-2009/10), causal peer effects during soccer games are identified. This unique identification strategy is applicable for other studies to analyze complex interactions without simplifying the structure of those interactions.