Friday, November 27, 2009

Monthly death cycles

Daily mortality counts fluctuate over the course of a calendar month. As has been documented by Phillips et al. (1999), daily mortality decreases to about one percent below the average in the week prior to the 1st day of the month, and then increases to almost one percent above the average in the first few days of the month. This within-month mortality cycle is particularly pronounced for homicides, suicides, and accidents. ...

Updating and extending the earlier work of Phillips et al., we document within-month mortality cycles for many causes of death, including external causes, heart disease, heart attack, and stroke (but not cancer). The within-month cycle is also evident for both sexes and for all age groups, races, marital status groups, and education groups.

[We] obtained daily data on a number of different activities and purchases, including going to the mall, visiting retail establishments, purchasing lottery tickets, going to the movies, and the amounts spent on food and non-food retail purchases. These data all show the same pattern, namely, that activity declines toward the end of the month and rebounds after the 1st of the month. ...

The concordance between the mortality and activity cycles leads us to conclude that an increase in activity leads to an increase in mortality. ...

We provide suggestive evidence that the rise in mortality is linked to changing liquidity over the month. First, we document that the peak-to-trough in mortality is greatest for those with low levels of education, a group that has been found to have liquidity problems. Second, we link liquidity to movements in consumption by showing there are smaller movements in activity and consumption over the month for groups we would expect to have less liquidity issues, namely, those in higher-income groups and those with more education. ... Finally, we provide direct evidence that mortality increases in the short term after the receipt of income. ...

First, seniors who enrolled in Social Security prior to May 1997 typically received their Social Security checks on the 3rd of the month. For this group, daily mortality declines just before paycheck receipt, and is highest the day after checks are received. Second, for those who enrolled in Social Security after April 1997, benefits are paid on either the second, third or fourth Wednesday of the month, depending on beneficiaries’ birth dates. Among this group, mortality is highest on the days checks arrive. Third, the Alaska Permanent Fund pays residents of Alaska an annual dividend, and during the week that direct deposits are made, mortality among urban Alaskans increases by 13 percent. Fourth, during the week the 2001 tax rebate checks arrived, mortality among 25-64 year olds increased by 2.5 percent. Finally, counties with a large percentage of their population in the active military experience relatively large spikes in mortality among 17-64 year olds immediately after the 1st and the 15th of the month, the dates on which military personnel are paid.
--William Evans and Timothy Moore, "Liquidity, Activity, and Mortality," on the upside of being short on cash

No comments: