Consumption and saving under liquidity constraints: The Philippine case, 1988-2000

• Main Author: See, Edward Camaligan
• Format: text
• Language: English
• Published: Animo Repository 2004
• Subjects: Consumption (Economics); Liquidity (Economics); Saving and investment; Saving and thrift; Economics
• Online Access: https://animorepository.dlsu.edu.ph/etd_masteral/3169; https://animorepository.dlsu.edu.ph/context/etd_masteral/article/10007/viewcontent/CDTG003685_P.pdf
• Institution: De La Salle University

This thesis explores the consumption-saving behavior of Filipino households under liquidity constraints. It raises the following questions: Is the consumption and saving behavior of Filipino households consistent with the predictions of Permanent Income Life Cycle Hypothesis (PILCH)? Are Filipino households liquidity-constrained (i.e., unable to draw on savings in illiquid assets or future incomes to finance current expenditures)? Which households are liquidity-constrained? (That is, which household characteristics are the correlates of liquidity-constrained status?) Do liquidity constraints affect the consumption and saving behavior of Filipino households? Do different demographic household characteristics produce different profiles of consumption, income, and saving over the life cycle? In answering the questions above, this study utilizes the cohort analysis using pseudo panels (synthetic panels). Data from the Family Income and Expenditures Survey of the years 1988, 1991, 1994, 1997, and 2000 are employed to generate the pseudo panels used in the construction of the age-consumption and age-savings profiles from a select set of the demographic characteristics of the household head, such as the gender, class of worker, and educational attainment of the household head, and from the variables related to the household, such as the type of household living arrangement and whether households receive transfers or not. Furthermore, the demographic characteristics mentioned are subdivided according to whether a household is liquidity constrained or unconstrained. The unconstrained and constrained samples are derived using the sample separation technique. The technique utilizes a probit model of liquidity constraint derived from the 1999 Annual Poverty Indicators Survey. A regression based on the Euler equation is used to verify whether liquidity constraint binds among Filipino households. This thesis contains the following findings: The whole sample age-expenditures profiles revealed a hump shape profile, which violated the PILCH. However, an interesting finding came out when the sample was divided. This highlights the role of the credit institutions in shaping the patterns of consumption and saving of households over the lifetime. As a way to minimize adverse selection problem among financial institutions, the practice of credit scoring and credit rationing are commonly observed. Liquidity constraints are results of the said behavior of financial institutions. Dividing the sample according to the liquidity constraints, the age-expenditures profile of the unconstrained group was observed to be flatter compared to hump shape profile of the constrained group. The analysis went further by controlling for the demographic characteristics related to the household and the household head. In the whole sample, hump shape age-expenditures profiles were observed in households who do not receive transfers, whose heads that are working in private establishments or are self-employed, whose heads have low educational attainment, whose family living arrangement is nuclear type, and whose heads are males. Turning to the result using the unconstrained sample, the age-expenditures profiles were seen to be flat as expected. However, one interesting finding surfaced. The age-expenditures profiles of households in nuclear family living arrangements remained hump shape. The effects of liquidity constraints on consumption appeared stronger compared to the effects of living in nuclear families. For the constrained sample, age-expenditures profiles were observed to be hump shape as expected. Two interesting contradictions appeared. The age-expenditures profiles of the households living in an extended family arrangement and the households xxii who receive transfers were observed to be flat regardless of whether they are constrained or not. A conclusion can be made out the results mentioned that strong family ties, altruistic behavior of the households, and reciprocity behavior are strong among Filipino households. The flat age-expenditures profiles were observed, arguably, through the internalization of social insurance and through inter and intra-household allocation of funds. Turning to the results of the regression, the liquidity constraint hypothesis emerged strong to be rejected. With the first model specification, the coefficients of the income variable in the three regression using whole, unconstrained, and constrained samples are statistically significant, supporting the excess sensitivity of consumption to the changes in income argument and the liquidity constraint hypothesis. The second model specification used interaction variables to have the effect of income controlling for the cohort effects. The coefficients of the interaction variables turned out to be statistically significant among the three sets of samples. However, different degrees of the change in the growth rate of expenditures were observed in the unconstrained and constrained groups. The change in the growth rate of expenditures explained by the changes in income is higher in the constrained sample. This supports the hypothesis that consumption is more sensitive to income changes in the constrained group. The results regarding characteristics related to the household and to the head yielded interesting findings. In the constrained sample, family size and the gender of the head appeared to be statistically significant. The insignificance of the rest of the demographic variables supports the theory that consumption growth is not affected by the changes of the demographic variables, but is affected by the changes in income (consumption tracking income). On the other hand, variables related to the characteristics of the household and the household head, such as family size, the number of household members 6 years old and below, the type of household living arrangements (nuclear and extended families), and educational attainment of the household head (elementary level/graduate, high school level/graduate, and undergraduate) appeared to be statistically significant. Again, this result supported the theory that consumption growth rate reflects the needs of the households in two periods (from current to the next) in the absence of credit