Campbell and Mankiw (1990) examine the consistency of the permanent income hypothesis with postwar U.S. aggregate data. They propose a model where a fraction of income, denoted λ, is consumed immediately, while the remainder is consumed based on permanent income. They estimate λ to be around 40-50%, indicating a significant deviation from the permanent income hypothesis. This finding is robust to various statistical issues, including time aggregation, and cannot be easily explained by changes in real interest rates or non-separabilities in the utility function. The paper presents an instrumental variables approach to test the permanent income hypothesis. It shows that a value of λ greater than zero can explain excess sensitivity of consumption to income and insufficient variability in saving. The authors use instrumental variables to estimate λ and test the hypothesis that λ = 0. Their test is valid regardless of whether income has a unit root and is more powerful than standard unrestricted tests for consumption following a random walk. By lagging instruments two periods, they avoid econometric difficulties caused by time aggregation. The authors also test their framework against more general time-series representations of consumption and income, such as error-correction models. They generalize the approach to handle alternative versions of the permanent income hypothesis, including changes in real interest rates and non-separabilities in the utility function. They examine whether these alternative formulations of preferences can explain the apparent excess sensitivity of consumption to income. The empirical results show that the permanent income hypothesis is strongly rejected. The estimated λ is around 0.4-0.6, indicating a substantial departure from the hypothesis. The results are robust to different specifications and data samples. The authors conclude that the permanent income hypothesis is not consistent with postwar U.S. data, and that consumption is more predictable than the hypothesis suggests. The findings suggest that a significant portion of income is consumed based on current income rather than permanent income.Campbell and Mankiw (1990) examine the consistency of the permanent income hypothesis with postwar U.S. aggregate data. They propose a model where a fraction of income, denoted λ, is consumed immediately, while the remainder is consumed based on permanent income. They estimate λ to be around 40-50%, indicating a significant deviation from the permanent income hypothesis. This finding is robust to various statistical issues, including time aggregation, and cannot be easily explained by changes in real interest rates or non-separabilities in the utility function. The paper presents an instrumental variables approach to test the permanent income hypothesis. It shows that a value of λ greater than zero can explain excess sensitivity of consumption to income and insufficient variability in saving. The authors use instrumental variables to estimate λ and test the hypothesis that λ = 0. Their test is valid regardless of whether income has a unit root and is more powerful than standard unrestricted tests for consumption following a random walk. By lagging instruments two periods, they avoid econometric difficulties caused by time aggregation. The authors also test their framework against more general time-series representations of consumption and income, such as error-correction models. They generalize the approach to handle alternative versions of the permanent income hypothesis, including changes in real interest rates and non-separabilities in the utility function. They examine whether these alternative formulations of preferences can explain the apparent excess sensitivity of consumption to income. The empirical results show that the permanent income hypothesis is strongly rejected. The estimated λ is around 0.4-0.6, indicating a substantial departure from the hypothesis. The results are robust to different specifications and data samples. The authors conclude that the permanent income hypothesis is not consistent with postwar U.S. data, and that consumption is more predictable than the hypothesis suggests. The findings suggest that a significant portion of income is consumed based on current income rather than permanent income.