SECCIÓN 501 LUGARES CLASE
B. Disposiciones específicas para los distintos tipos de transformador.
2) Según lo que establece el Artículo 450-
The changes in normal retirement age and delayed retirement credit impact the retirement decision only in the life-cycle model through a change in the SSW. But, the earnings test can affect the retirement decision in the life-cycle context only under conditions that make the actuarial adjustments less than fair and if labor market rigidities restrict the choices available to an individual. The approach chosen by previous
researchers studying the response of work decision among older individuals to earnings test changes depends on whether they model the earnings test as a tax in the life-cycle or
static context.23 I attempt to assess the impact of the earnings test on the retirement decision of individuals in a life-cycle context.
Researchers in the 1980s studied the impact of increases in SSW on the labor force participation of older men using time series variation in benefits. These increases in SSW arose either due to ad hoc changes implemented in the late 1960s and early 1970s or due to overindexation of benefits in the 1970s. The findings from these studies support the view that higher social security benefits lower labor force participation among older men.24 Krueger and Pischke (1992) note the difficulty in separately identifying the
impact of social security benefit increases in time-series analysis from other variables that have trended over time. They instead identify the effect of social security benefits by relying on an exogenous unanticipated downward movement in benefits as a result of the 1977 Amendment which permanently lowered the benefit level for a cohort of individuals termed the “notch babies”. They use aggregated CPS data from 1976-88. In their analysis they control for actuarial adjustments applied to foregone benefits by including a measure for accrual of benefits. They assess the impact of social security benefit reductions on the labor supply of men by examining the response along three dimensions: labor force participation, self-reported retirement, and the number of weeks worked last calendar year. They find no robust significant influence of social security benefit reductions on the retirement decision.25
23 As described in the previous section, given the recent increases in the DRC, at an actuarially
fair level of DRC, the earnings test is not a tax in the life-cycle model. But, if individuals are unaware of these credits, face borrowing constraints, or have lower than average life expectancy then the earnings test may still be a tax in the life-cycle context.
24 Krueger and Pischke (1992) discuss the methodology used in these studies.
25 Moulton and Stevens (2015) emphasize that the Notch legislation changed both retirement
The finding of Krueger and Pischke’s study analyzing the effect of benefit reductions cannot be directly applied in evaluating the response in retirement age due to the recent benefit reductions. The recent benefit reductions are instituted through a change in the normal retirement age instead of a permanent benefit cut with an unchanging normal retirement age. In light of past evidence of large spikes in the retirement hazard at the social security early and normal retirement age, and recent evidence of a shift in these spikes it may be important to control for any noneconomic norm related effects that may impact the retirement decision in addition to the financial incentives. This norm related effect may arise if older individuals view the normal retirement age as a focal point (guidance) for retirement.
Gruber and Orszag (2003) use aggregate CPS data from 1973-98 to examine the effect of earnings test policy changes on the work decision, hours worked, and claiming behavior of older men and women.26They also evaluate the impact of the 1983 earnings test repeal for individuals between ages 70-72. In their study they identify the ad hoc increases implemented from 1996 onwards in the earnings test threshold amount as an important source of variation in earnings test parameters among individuals for whom the earnings test was still in place. But, they find no robust evidence of a change in work decision of older men and women in response to a loosening of the threshold amount.
legislation reduced incentives for delaying retirement as it changed the benefit computation formula to not include earnings beyond age 61; it is, thus, inaccurate to interpret Kruger and Pischke’s findings as evidence of insensitivity of retirement to wealth levels.
26 Some other studies that have analyzed the impact of the earnings test on work decision are as
They, however, find robust evidence that the threshold amount increases lead to higher benefit receipt among both men and women.
In their analysis Gruber and Orszag note that their finding of no impact of earnings test parameters on the work decision of older workers is best understood when applied to individuals facing the earnings test. Due to the relatively fewer observations available for individuals between ages 70-72, they are unable to precisely measure the full impact of the earnings test repeal. They also ignore the effect of the delayed
retirement credit under the assumption that older workers either are misinformed, or do not understand delayed retirement credit adjustments, and hence the earnings test acts as a tax for all individuals. This view may be correct for the time period covered in their study, but cohorts reaching normal retirement age in 2002 and later have a delayed retirement credit of 6.5 percent per year or higher (which is approximately actuarially fair for the average individual). The birth cohorts affected by the recent earnings test repeal of 2000 experienced increases in their normal retirement age, and/or delayed retirement credit. Keeping in mind the recent increases in the delayed retirement credit and normal retirement age (which did not affect the cohorts studied in Gruber and Orszag’s analysis), it seems important to try and assess how changes in earnings test parameters affect individuals’ retirement decision in a life-cycle context controlling for the influence of DRC and NRA changes.
Three recent studies have examined the effect of both changes in normal retirement age and changes in the delayed retirement credit on the labor force
participation or employment decision of older men; two of these studies also control for the effect of the 2000 earnings test repeal. Blau and Goodstein (2010) create a synthetic
panel data using CPS and SIPP which is aggregated by birth cohort, year of age and four education groups covering for the time period 1962-2005. They use Social Security Administrative (SSA) data to match the average lifetime earnings for each cohort which are then used to derive the average social security benefits for each cohort. Blau and Goodstein’s specification only accounts for financial incentive effects of NRA changes, but it does not allow them to capture the norm effects. They ignore the earnings test noting that there is no straightforward way for measuring its effect in their framework. They find the changes in normal retirement age and delayed retirement credit explain about one fourth to one half of the recent increase in the labor force participation of older men.
Mastrobuoni (2009) focuses on the effect of changes in normal retirement age. He uses CPS data from 1989-2007 to identify the impact of NRA increases on the retirement age of both men and women by using a regression discontinuity design, and he restricts his analysis to individuals between ages 61-65. He identifies the impact of the normal retirement age by estimating the difference between the yearly trend in average
retirement age of cohorts born before 1938 (not affected by the NRA increases) relative to cohorts born after 1937, while controlling for the influence of the DRC changes and earnings test repeal. He finds that a 2-month increase in the normal retirement age increases the average retirement age of the affected cohorts by 1-month.
Pingle (2006) focuses on the effect of DRC changes. Using data from SIPP panels covering a period of 1983-2003, he finds that a delayed retirement credit increase of 1 percent raises the employment rate of 65-70-year-old men by 1.5 percent. Pingle’s analysis extends only until 2003 when individuals in the first cohort affected by the NRA
increases attain their normal retirement age, so he is unable to assess the impact of normal retirement age changes among those above their NRA. Both Mastrobuoni and Pingle’s study control for effect of the earnings test repeal on labor force participation. Their specifications allow them to capture the norm related effects of normal retirement age changes because they capture these changes by including a variable indicating the change in the NRA assigned by birth year instead of a variable for changes in social security benefits. But they are unable to separately identify the norm and financial effects associated with NRA changes.
The three studies that analyze the recent policy changes can be improved upon in at least four ways. First, none of these studies incorporate the sizeable variation in the earnings test threshold amount (noted by Gruber and Orszag) that were introduced through the ad hoc changes implemented after 1996. Second, even though older women are a sizeable part of the labor force there is only one study (Mastrobuoni, 2009)
analyzing the impact of recent changes in the NRA and DRC on the retirement decision among older women between ages 61-65, and none examining the retirement decision changes of women above their normal retirement age. Third, there is little work in the recent research disentangling the financial and norm related effects arising from the NRA changes. It is possible to separately identify these effects through use of a dataset that includes accurate birth year information of individuals. Finally, previous researchers have only focused on changes in the labor force participation decision of older workers in response to the recent changes in Social security policy. Given the large evidence of the variety of options explored by older individuals in their transition to retirement, it seems important to analyze how older workers response differs when evaluated using alternative
definitions of retirement. I attempt to include these suggested changes to the previous work in my empirical analysis.