Printer Friendly

Income mobility and the Earned Income Tax Credit.

I. INTRODUCTION

According to the Committee on Ways and Means (1993, 1052) the original purposes of the Earned Income Tax Credit (EITC) were to target tax relief to low-income families with children, offset Social Security taxes to these families, and improve incentives to work. The EITC, which began in 1975 as a minor program, has grown to become a central part of the federal government's antipoverty program. (1) Technically a refundable tax credit, the EITC provides an income transfer to low earning families. (2)

The EITC is made up of three ranges: the phase-in range, the maximum credit range, and the phase-out range. In the phase-in range, the EITC acts as a wage subsidy; as the family earns more the transfer increases. In the maximum credit range, the EITC acts as a lump-sum subsidy; the transfer is the same regardless of earnings. In the phase-out range, the EITC acts as a negative income tax; as the family earns more, the tax credit is reduced. Both lump-sum subsidies and negative income taxes discourage work effort. (3)

The EITC as a whole and especially the wage subsidy in the phase-in range have received wide support from economists. (4) However, the upper limit on the phase in range is so low that it is unlikely that the program has much positive work incentive because most people will end up in either the flat or the phase-out ranges, both of which have work disincentives. (5) For example, a person working full-time at a minimum wage job will earn $10,300 ($5.15 * 40 hrs * 50 weeks), which is more than the 2001 phase-in threshold of $10,020 for families with two or more children.

This article has three purposes. The first is to calculate the dynamics of EITC receipt. Previous studies have not discussed the dynamics of eligibility or receipt for the EITC. (6) Examining the dynamics of participation and eligibility for income transfer programs, such as the EITC, is critical to obtain a full understanding of what the programs do and who they serve. This article helps close this gap.

My second purpose is to determine if the importance of the phase-in range is exaggerated by annual data. Annual data shows that about 27% of EITC recipients are in the phase-in range in a given year. (7) However, as discussed, the upper limit on the phase-in range is so low that most people will end up in the maximum credit or the phase-out range, both of which have work disincentives.

The third purpose of this article is to determine why families move in and out of the EITC. Do families move in and out because of changes in the number of children, changes in family structure (such as the loss or gain of a spouse), or some other reason?

II. CHANGES OVER TIME IN THE EITC PARAMETERS

The parameters used in this article were taken from the Committee on Ways and Means, Overview of Entitlement Programs (1998). Table 1 shows the EITC parameters from 1975 to 2001. (8) As mentioned above, the EITC is made up of three ranges: the phase-in range, the maximum credit range, and the phase-out range. For example, in the 1990 phase-in range, eligible families could have received a wage subsidy of 14% (column [2]) of their earnings up to $6,810 (column [3]). This means that in the phase-in range, when a family earned an extra $1000 they could receive a tax credit of $140 (0.14 * 1000). (9) In the 1990 maximum credit range, families with earnings between $6810 (column [3]) and $10,730 (column [7]) could have received a lump sum-subsidy of $953 (column [5]). (10) In the 1990 phase-out range, the EITC was phased out at a rate of 10% (column [6]) until the benefit was zero at $20,264 (column [8]). (11)

Column (4) of Table 1 shows how much the family would earn if they had one person working full time (2000 hours a year) and earning the minimum wage. If only one spouse earned the minimum wage and worked full-time for a year, the family earned too much to be in the phase-in range. (12) Families in the phase-in range work only part-time.

Table 1 also shows how the EITC parameters changed over time. In 1979, 1985, 1986, 1990, and 1993, the federal government enacted legislation that expanded the EITC. (13) The 1979 expansion introduced the maximum credit range. Though the 1979 and 1985 laws expanded the EITC, the expansions were not large enough for the credit to keep up with inflation.

In 1986, 1990, and again in 1993, Congress substantially expanded the credit. (14) The Tax Reform Act of 1986 increased the credit amounts and indexed them for inflation. The budget act of 1990 (OBRA, 1990) increased the credit amounts over several years and taxpayers with more than one child started to receive a slightly larger credit than taxpayers with one child. OBRA 1990 also added auxiliary credits for very young children and for health insurance premiums paid on behalf of a qualifying child. However, these auxiliary credits were repealed in 1993. Also in 1993, eligibility for the credit was expanded to include childless families.

III. THE PSID DATA AND THE DISTRIBUTION OF RECIPIENTS

The income, earnings, and demographic data used herein were taken from the Panel Study of Income Dynamics (PSID; 1992) for the years 1975-92. (15) There are 9148 families in the data set. Only families who had earnings of at least $100 in annual labor income and had one or more eligible children were included when calculating the tax credit. (16) Because the PSID oversamples the poor, the data were weighted using the family weights.

The PSID does not report if a family actually received the EITC. However, the PSID can be used to estimate if a family is eligible to receive the EITC. Some eligible families may choose not to receive EITC benefits. Hoffman and Seidman (1990) found that 10.7% of all families in the United States received EITC benefits in 1988. For comparison, in 1992, 7.93% of all PSID families were eligible for the EITC. (17)

As shown in Table 2 column (2), the GAO (1996) found that in 1994 14.9 million EITC recipients had at least one qualifying child. (18) Of these 14.9 million families, 4 million were in that phase-in range, 2.1 million were in the maximum credit range, and 8.9 million were in the phase-out range. As shown in column (3), most recipients (59.5%) are in the phase-out range.

The fourth column in Table 2 reports the percentage of PSID families in each income range from among those who were eligible in 1992. About 31% of these families were eligible for the phase-in range, 15.26% were eligible for the maximum credit range, and 53.80% were eligible for the phase-out range. This compares favorably with the GAO's (1996) calculations (column [3]) that 26.6% 01 actual recipients were Hi the phase in range, 13.9% of actual recipients were in the maximum credit range, and the 59.5% were in the phase-out range in 1994.

IV. DESCRIPTION OF SPELL LENGTH

Table 3 describes the length of spells on the EITC. (19) A spell is the amount of time that a family is eligible to receive the EITC. (20) The beginning of a spell is the first year of eligibility and the ending of the spell is the last year of eligibility. (21) Because the EITC began in 1975 and the data goes until 1992, families can be eligible to receive EITC benefits for a maximum of 18 years. Families can have several spells spread over the 18 years of data. Column (1) of Table 3 shows the length of each spell in years.

Bane and Ellwood (1983, 1994) use the example of a hospital to illustrate the difference between families beginning a spell and families at a point in time. Most people admitted to a hospital will only require short spells of hospitalization. But a few people are chronically ill and will require long periods of hospitalization. Even though chronically ill people are a small part of new spells, because they stay in the hospital longer they end up being a larger fraction of the hospital's population at a point in time. (22)

D(t), in column (2), is the fraction of families who are eligible to have EITC spells which last exactly t years. D(t) is analogous to measuring the time each patient spends in the hospital and then calculating how many are patients for one day, how many are patients for two days, and so on. (23)

Column (2) shows that most EITC spells are relatively short. Fifty-one percent of spells last one year and 74% of spells are over in two years or less. Only 9% of spells last five years or longer. The average spell length is 2.135 years.

The exit probability p(t) in column (3) is the probability that a household who is in their tth year of EITC eligibility ends their spell at the end of the year. p(t) is analogous to determining the probability that a patient who has been in the hospital for t days will be released at the end of the day. Equation (1) calculates p(t). (24)

(1) p(t)=D(t)/[summation over[t-1]](j=1)D(j)].

In equation (1), the denominator is the fraction of households whose EITC eligibility survives from year 1 until year t-1. In year 1, p(l) = D(1) where D(1) is the 51% of households whose spell lasts exactly one year and p(1) is the probability that households in the first year of a spell are not eligible to receive the EITC the second year. Forty-eight percent of households who are eligible in year 2 will not be eligible the third year, that is, p(2) = D(2)/[1 -D(1)].

Column (4) shows F(t), which is the fraction of all households eligible to receive the EITC at a point in time who will be eligible for exactly t years. F(t) is analogous to taking a census of patients on a certain date and determining the total time they stayed in the hospital. Those that stayed in the hospital one day would be listed in row 1, those that stayed in the hospital for two days would be listed in row 2, and so on. F(t) is calculated by equation (2).

(2) F(t) = [t (D(t)]/[[summation over ([infinity]/j=1)] jD(j)].

For simplicity, equation (2) assumes that there is a no growth steady state. (25) So the number of families starting spells in each year is constant. The average spell length for families at a point in time is 3.55 years (column [4]). This is nearly 1.5 years longer than the 2.13-year average spell length for families just starting a spell. However, even at a point in time, spell length is still relatively short. Sixty-one percent of households eligible at a point in time are in spells that last three years or less. Ninety-one percent are over in 7 years or less.

Column (5) shows G(t), the fraction of all persons eligible at a point in time who have been recipients for t years. In other words, the fifth column is the distribution of households who are eligible to receive the EITC at a point in time but have not completed their spells. In the hospital example, G(t) would be a census of how long the current patients have been in the hospital. G(t) is calculated by equation (3).

(3) G(t) = [1 - [summation over (t-1/j=1)] D(j)]/([summation over ([infinity]/s=1)][1 - [summation over (t-1/k=1)] D(k)]).

Equation (3) calculates the fraction of families who began spells t years earlier who would still be on the program and renormalizing assuming a no-growth steady state.

The problem with G(t) is that some of these patients who just entered the hospital will actually be long-term patients. In other words, G(t) overstates mobility. G(t) makes it look like spells will be shorter than they really are. Column (4) gives a more realistic view of spell duration at a point in time.

Families eligible to receive EITC were much more mobile than families who were eligible to receive Aid for Families with Dependent Children (AFDC). For example, Bane and Ellwood (1983, 1994) found that the average AFDC spell was 4.7 years for families just starting a spell and 11.1 years for the completed spell distribution. In contrast, the average EITC spell was 2.1346 years for families just starting a spell (column [2]) and 3.55 years for the completed spell distribution (column [4]). In other words, families who just became eligible for the AFDC on average received AFDC benefits for 2.5 years longer than families who just became eligible for the EITC. Likewise, looking at a point in time, AFDC families received benefits an average of 7.5 years longer than families eligible to receive the EITC. These numbers mean families who are eligible for the EITC are eligible to receive benefits for a relatively short period of time relative to the AFDC program where a few families receive benefits for a relatively l ong time.

There are two potential problems with Table 3. First, because the EITC was greatly expanded in the 1980s and 1990s, the estimates may overstate mobility. There should be more mobility when EITC parameters are more narrow. Second, a sizable portion of the families who began a spell in 1992 or before had not completed it by the end of 1992. In other words, this data is right-censored.

To see the effect of changing program parameters and spells that have not been completed, Table 4 keeps program parameters constant and only includes completed spells. Program parameters were held constant by using 1992 parameters to calculate EITC benefits for each year. The Consumer Price Index (CPI) was used to adjust dollar amounts for inflation. Because of using the 1992 parameters, the percent of households that were eligible to receive EITC benefits for at least 1 year during the 18-year period increased from 24.05% to 27.46%.

Using 1992 parameters and only including spells that have been completed by 1992 causes the average spell length to increase from 2.135 years in Table 3 to 2.219 in Table 4. In other words, the average spell length is about a month longer. The completed spell distribution for families at a point in time (Table 4, column [4]) was 4.087 using completed spells and 1992 parameters. This is about six months longer than the 3.554 average spell length in Table 3. Changing parameters and right-censored data doesn't seem to have a large effect on the distribution. (26)

V. RECIDIVISM

The data reported in Tables 3 and 4 are for spells. However, spells don't take into account total time on the EITC. Many households who stop being eligible are not permanently ineligible to receive the EITC. The recidivism rate (R(t)) in column (2) of Table 5 is the probability that a household who is in their tth year of not being eligible starts a new spell at the end of the year. For example, approximately 16% of households became eligible to receive EITC benefits again after only one year of ineligibility. Of those households who were not eligible for two years, about 31% became eligible to receive EITC benefits the third year.

Column (3) shows E(t), which is the percentage of households who after the tth year are again eligible to receive the EITC. About 16% of households are eligible to receive the EITC after exactly one year of ineligibility, and about 7% of households are eligible to receive the EITC after exactly two years of ineligibility. Over the 17 years, 38.7% of households eventually return to EITC eligibility. However, all of these households returned in the first 14 years. E(t) is calculated as

(4) E(t) = [1 - [summation over (t-1/j=1)] E(j)]R(t).

Column (4) shows the cumulative return of households eligible to receive the EITC. As mentioned above, 38.7% of households eventually return to EITC eligibility. Note that 33% of households return in the first five years. Only 5.7% return after five years.

As before, there are two potential problems with Table 5. First, the EITC was greatly expanded in the 1980s and 1990s, and second, a sizable portion of the families who began a spell in 1992 or before had not completed it by the end of 1992. To see the effect of these two potential problems, Table 6 keeps program parameters constant and only includes completed spells. As before, 1992 parameters were used to calculate EITC benefits for each year, and dollar amounts were adjusted for inflation using the CPI.

Including only completed spells and using the 1992 parameters does not have much effect on the results. The recidivism rate increased from 38.7% to 38.9%. Instead of 33% of households returning to eligibility within five years of losing eligibility, now 35% of households return to eligibility within five years.

VI. DESCRIPTION OF SPELL LENGTH IN THE PHASE-IN AND PHASE-OUT RANGES

As was discussed, the second purpose of this article is to determine if the importance of the phase-in range is exaggerated by annual data. During the time of this study, a family with a member who works full-time at minimum wage earned too much to be in the phase-in range. Also, in the phase-in range, the EITC is a wage subsidy, giving families an incentive to work more. Working more means they are likely to earn more than the phase-in range. Because of these two effects, it is likely that families who are eligible to be in the phase-in range will be more mobile than families who are eligible to be in the phase-out and maximum credit ranges, where the EITC reduces work incentives and act like a negative income tax. (27)

Table 7 compares the distribution of spells for families who were in the phase-in range and families who were either in the phase-out or maximum credit ranges. The phase-in range was treated as one program and the combined maximum credit/phase-out range was treated as another program. For example, if a household is in the phase-in range and then moves to the phase-out range, it is the end of a spell for the phase-in range and the beginning of a spell in the combined maximum credit/phase-out range. Parameters from 1992 are used to calculate the results in Table 7.

According to Table 7, families beginning a spell in the phase-in range are slightly more mobile than households in the combined maximum credit/phase-out range. The average spell length is about six months longer for the maximum credit/phase-out range than for the phase-in range (2.059 years versus 1.553 years).

Households in both the phase-in range and the maximum credit/phase-out range are quite mobile. For example, 67% (column [2]) of phase-in families and 55% (column [5]) of maximum credit/phase-out range families will have a spell that lasts only one year.

It seems plausible that the EITC's financial incentives would have little effect on families who are only eligible for the EITC for a short period of time. For example, the work disincentives in the phase-out and maximum credit ranges may have little or no effect on families who are only briefly eligible because of unemployment, illness, family dissolution, or some other reason. Likewise, the work incentives in the phase-in range may have little or no effect for those families that are only briefly eligible. Although somewhat arbitrary, it seems preferable to focus on those who have spells of at least three years. These families can be thought of as normally having very low earnings and for them the financial incentives are more meaningful.

Approximately 27% of the EITC recipients were in the phase-in range in 1994. This suggests that only about 27% of the EITC recipients may realize an increased incentive to work. About 29.5% (column [4]) of these phase-in range households remained in the phase-in range for at least three years. This implies that only about 8% (27% * 29.5%) of all households eligible to receive the EITC remained in the phase-in range for at least three years. If three years is the relevant time period, then only about 8% of the EITC recipients may realize an increased incentive to work.

Because most households are in the phase-out and maximum credit ranges, the EITC mainly acts as a negative income tax that is not paid to those who don't work. This implies that on average the EITC reduces the recipients incentives to work. However, even though about 73% of recipients were in the phase-out and maximum credit ranges in 1994, only 51.4% (column [7]) of eligible families remained in these ranges for at least three years. This implies that only 38% (73% * 51.4%) of all EITC recipients were in the phase-out and maximum credit ranges for at least three years.

Note that in the annual data, the number of households in the phase-out and maximum credit ranges (73%) is about three times as many as in the phase-in range (27%), but when one compares households who typically have incomes in these ranges, there are almost five times as many in the phase-out and maximum credit ranges (38%) as there are in the phase-in range (8%). This suggests that the families subjected to meaningful disincentives may be an even larger part of the overall picture.

However, because the EITC is not paid to people who don't work, it may increase people's incentives to become employed and once they are employed to stay employed. For example, Eissa and Liebman (1996) found a 2.8% increase in labor force participation for single women with children. However, as will be shown in section VIII, families becoming eligible to receive the EITC because they were earning more than the EITC and then their earnings dropped is about twice as likely as families becoming eligible for the EITC because they were earning less than the EITC and their earnings increased.

Table 8 shows that heads of families starting a spell in the phase-out range were more likely to be working than heads starting a spell in the phase-in range. For example, 82.3% of heads who were eligible for the phase-out range reported that they were working, whereas about 51% of heads eligible for the phase-in range reported that they were working.

A family with a single earner may earn enough to be in the phase-in range but also report that he or she was not working. (28) This is because earnings are recorded the year after they were earned and not working is recorded at the time of the interview. For example, suppose a female head earned $5000 in 1990 and worked for seven months. If the 1990 interview was during one of the five months she was not working, then she would be listed as not working and she would be eligible for the phase-in range.

Heads in the phase-in range on average worked a little less than half time (920 hours a year). On the other hand, heads in the phase-out range worked an average of 1782 hours a year, which is nearly full-time.

VII. WHY HOUSEHOLDS ENTER AND LEAVE THE EITC

The third purpose of this article is to determine why families move in and out of the EITC. Table 9, column (1) shows six hierarchical reasons why households became eligible for an EITC spell. (29)

1. The procedure was first to look for families whose wife became the head. The wife could become the head either the year before a spell started or the year that the spell started.

2. If the wife did not become the head, then it was determined if a spell started when the EITC began in 1975. If the spell started in 1975, then entered sample qualifying in 1975 was classified as the reason for beginning a spell.

3. If a spell did not start in 1975, then it was determined if the family had at least one eligible child the year the spell started but no eligible children the year before. In this case, family now has eligible child was classified as the reason for beginning a spell.

4. If none of the first three reasons were true, it was determined if a family became eligible because its' earnings decreased. These are families who earned more than the EITC but now are eligible because their earnings fell.

5. If the first four reasons were not true, it was determined if a spell started because the family was earning too little to be eligible for the EITC but became eligible because the family started earning more. If this is the case, then higher earnings was classified as the reason for starting the spell.

6. If none of the first five reasons were true, then the family is classified as all other.

Column (2) shows the percentage of households that became eligible for each of the six reasons. The main reasons for becoming eligible to receive the EITC are (1) lower earnings (52.62%), (2) higher earnings (18.89%), and (3) family now has eligible child (11.83%). About 83.34% of all households became eligible by these three reasons.

Earning less (52.62%) caused more than twice as many spells as earning more (18.89%). However, these data do not answer the question if households who earned more than the EITC worked fewer hours to become eligible for the EITC or if households who were not working started working to become eligible.

Column (3) shows the reasons why families became eligible for the phase-in range. About 57% of families became eligible because they had lower earnings. These families were in the phase-out range, were in the maximum credit range, or earned too much to be eligible for the EITC. The families then earned less and became eligible for the phase-in range. The second largest reason to become eligible was higher earnings (23.49%). These are families that earned less than $100 a year and then earned enough to be in the phase-in range.

Column (4) shows the reasons why families became eligible for the combined maximum credit/phase-out range. The 46.59% of families who became eligible because of lower earnings were families who earned more than the EITC and then became eligible after they earned less. On the other hand, the 26.80% of families who became eligible after they earned more either were not working or were in the phase-in range but then earned enough to be in the combined maximum credit/phase-out range.

Column (1) in Table 10 shows six hierarchical reasons why families stopped being eligible for the EITC.

1. First, it was determined if the head married or wife remarried either the year before or the year after the spell ended. If the head married or the wife remarried, then head married or wife remarried was classified as the reason for ending eligibility for a spell.

2. If the head didn't marry or the wife didn't remarry, then it was determined if the family exited the sample qualifying when the data ended in 1992. If the family was still eligible in 1992, then exited sample qualifying in 1992 was coded as the reason for exiting the spell.

3. If the spell ended before 1992, then it was determined if the spell ended because the last eligible child left the household. If the last eligible child left, then last eligible child left is the reason for exiting the spell.

4. If none of the first three reasons were true and the family earns to much to be eligible for the EITC the year after the spell, then higher earnings is the reason for exiting a spell.

5. If none of the first four reasons were true and the family earns too little to be eligible for the EITC the year after the spell, then lower earnings is the reason for exiting a spell.

6. If none of the first five reasons were true, then the exit is classified as all other.

Column (2) shows the percentage of households that lose eligibility for each of the six reasons. The three most important reasons for exiting a spell are (1) families earnings increased above the EITC ranges (43.08%), (2) the household exited the sample qualifying in 1992 (18.79%), and (3) lower earnings (15.38%). More than twice as many families exited spells because of higher earnings (43.08%) relative to families who exited a spell because they dropped out of the labor force (15.38%).

Column (3) shows the percentage of households that became ineligible to receive the phase-in range of the EITC. The major reason for families to exit a spell in the phase-in range was higher earnings (46.80%). Higher earnings could be because the family earned enough to be in maximum credit range, in the phase-out range, or too much to be in the EITC. The second largest reason for exiting a spell is lower earnings (19.72%). These are families who lose eligibility because they earned less than $100 the next year.

Column (4) shows the percentage of households that lost their eligibility to be in either the phase-out or maximum credit ranges. The two largest reasons for families to exit a spell in the phase-out range were, first, higher earnings (43.86%) and, second, lower earnings (20.71%). Lower earnings includes families who earned less than the maximum credit range. These data imply that households gain and lose EITC eligibility because of changes in earnings rather than because of changes in family structure.

VIII. SUMMARY AND CONCLUSION

Households who are eligible to receive the EITC are mobile. Most spells are relatively short. For those families starting a spell, 51% of spells lasted one year and 74% of spells were over in two years or less. Only 9% of spells lasted five years or longer. Looking at families who are beginning a spell, the average spell length was 2.135 years.

Even for those families who have already begun a spell, most spells are relatively short. The average spell length was 3.55 years, which is about 1.5 years longer than for families just beginning a spell. However, 91% of spells were over in seven years or less, and 46% of spells were over in two years or less.

This article also finds that families eligible to receive EITC were much more mobile than families who were eligible to receive AFDC. For example, Bane and Ellwood (1983, 1994) found that the average AFDC spell was 4.7 years for families just starting a spell and 11.1 years for the completed spell distribution. In contrast, the average EITC spell was 2.1346 years for families just starting a spell and 3.55 years for the completed spell distribution. In other words, families who just became eligible for the AFDC on average received AFDC benefits for 2.5 years longer than families who just became eligible for the EITC. Likewise, looking at a point in time, AFDC families received benefits an average of 7.5 years longer than families eligible to receive the EITC.

About 72% of households became eligible to receive the EITC either because they earned more than the EITC but then had lower earnings (52.62%) or because they earned less than $100 a year and then earned more (18.89%). On the other hand, 16.46% of families became eligible after the wife became head (4.63%) or the family had an eligible child (11.83%). These data imply that over four times (71%/16%) as many households gain EITC eligibility because of changes in earnings rather than because of changes in family structure.

About 58% of families exited spells because of higher earnings (43.08%) or because they dropped out of the labor force (15.38%). On the other hand, about 23% exited a spell because either the last eligible child left (15.3%) or the head got married (7.36%). These data imply that over 2.5 times (48%/23 %) as many households lose EITC eligibility because of changes in earnings rather than because of changes in family structure.

Approximately 27% of the EITC recipients were in the phase-in range in 1994. This suggests that only about 27% of the EITC recipients may realize an increased incentive to work. About 30% of these phase-in range households remained in the phase-in range for at least three years. This implies that only about 8% (27% * 30%) of all households eligible to receive the EITC remained in the phase-in range for at least three years. If three years is the relevant time period, then only about 8% of the EITC recipients may realize an increased incentive to work.

Because most households are in the phase-out and maximum credit ranges, the EITC mainly acts as a negative income tax that is not paid to those who don't work. This implies that on average the EITC reduces the recipients incentive to work. For example, in the annual data, the number of households in the phase-out and maximum credit ranges (73%) is about three times as many as in the phase-in range (27%). When families who have spells of at least three years are considered, the families subjected to meaningful work disincentives may be an even larger part of the overall picture. For example, when you compare households who have incomes in these ranges for at least three years, there are almost five times as many families in the phase-out and maximum credit ranges (38%) as there are in the phase-in range (8%).

However, because the EITC is not paid to people who don't work, it may increase people's incentives to become employed and once they are employed to stay employed. For example, Eissa and Liebman (1996) found a 2.8% increase in labor force participation for single women with children. However, families becoming eligible to receive the EITC because they were earning more than the EITC and then their earnings dropped is about twice as likely as families becoming eligible for the EITC because they were earning less than the EITC and their earnings increased.
TABLE 1

EITC Parameters, 1975-2001

 Minimum
 Credit Income Min. Phase-out
Calendar Rate for Max. Wage Max. Rate
Year (%) Credit Earnings * Credit (%)
(1) (2) (3) (4) (5) (6)

1975 10 $4,000 $4,200 $400 10.00
1976-77 10 4,000 4,600 400 10.00
1978 10 4,000 5,300 400 10.00
1979 10 5,000 5,800 500 12.50
1980 10 5,000 6,200 500 12.50
1981-84 10 5,000 6,700 500 12.50
1985-86 11 5,000 6,700 550 12.22
1987 14 6,080 6,700 851 10.00
1988 14 6,240 6,700 874 10.00
1989 14 6,500 6,700 910 10.00
1990 14 6,810 7,600 953 10.00
1991 (1 QC) 16.7 7,140 8,500 1,192 11.93
 (2 QC) 17.3 7,140 8,500 1,235 12.36
1992 (1 QC) 17.6 7,520 8,500 1,324 12.57
 (2 QC) 18.4 7,520 8,500 1,384 13.14
1993 (1 QC) 18.5 7,750 8,500 1,434 13.21
 (2 QC) 19.5 7,750 8,500 1,511 13.93
1994 (0 QC) 7.65 4,000 8,500 306 7.65
 (1 QC) 26.3 7,750 8,500 2,038 15.98
 (2 QC) 30.0 8,425 8,500 2,528 17.68
1995 (0 QC) 7.65 4,100 8,500 314 7.65
 (1 QC) 34.0 6,160 8,500 2,094 15.98
 (2 QC) 36.0 8,640 8,500 3,110 20.22
1996 (0 QC) 7.65 4,220 8,500 323 7.65
 (1 QC) 34.0 6,330 8,500 2,152 15.98
 (2 QC) 40.0 8,890 8,500 3,556 21.06
1997 (0 QC) 7.65 4,340 9,500 332 7.65
 (1 QC) 34.0 6,500 9,500 2,210 15.98
 (2 QC) 40.0 9,140 9,500 3,656 21.06
1998 (0 QC) 7.65 4,460 10,300 341 7.65
 (1 QC) 34.00 6,680 10,300 2,271 15.98
 (2 QC) 40.00 9,390 10,300 3,756 21.06
1999 (0 QC) 7.65 4,530 10,300 347 7.65
 (1 QC) 34.00 6,800 10,300 2,312 15.98
 (2 QC) 40.00 9,540 10,300 3,816 21.06
2000 (0 QC) 7.65 4,610 10,300 353 7.65
 (1 QC) 34.00 6,920 10,300 2,353 15.98
 (2 QC) 40.00 9,720 10,300 3,888 21.06
2001 (0 QC) 7.65 4,760 10,300 364 7.65
 (1 QC) 34.00 7,140 10,300 2,428 15.98
 (2 QC) 40.00 10,020 10,300 4,008 21.06


 Phase-out Range Phase-out Range
Calendar Beginning Ending
Year Income Income
(1) (7) (8)

1975 $4,000 $8,000
1976-77 4,000 8,000
1978 4,000 8,000
1979 6,000 10,000
1980 6,000 10,000
1981-84 6,000 10,000
1985-86 6,500 11,000
1987 6,920 15,432
1988 9,840 18,576
1989 10,240 19,340
1990 10,730 20,264
1991 (1 QC) 11,250 21,250
 (2 QC) 11,250 21,250
1992 (1 QC) 11,840 22,370
 (2 QC) 11,840 22,370
1993 (1 QC) 12,200 23,050
 (2 QC) 12,200 23,050
1994 (0 QC) 5,000 9,000
 (1 QC) 11,000 23,755
 (2 QC) 11,000 25,296
1995 (0 QC) 5,130 9,230
 (1 QC) 11,290 24,396
 (2 QC) 11,290 26,673
1996 (0 QC) 5,280 9,500
 (1 QC) 11,610 25,078
 (2 QC) 11,610 28,495
1997 (0 QC) 5,430 9,770
 (1 QC) 11,930 25,750
 (2 QC) 11,930 29,290
1998 (0 QC) 5,570 10,030
 (1 QC) 12,260 26,473
 (2 QC) 12,260 30,095
1999 (0 QC) 5,670 10,200
 (1 QC) 12,460 26,928
 (2 QC) 12,460 30,580
2000 (0 QC) 5,770 10,380
 (1 QC) 12,690 27,413
 (2 QC) 12,690 31,152
2001 (0 QC) 5,950 10,710
 (1 QC) 13,090 28,281
 (2 QC) 13,090 32,121

* Minimum wage earnings = Minimum wage x 2000 hours.

QC = Qualifying child.

Sources: Committee on Ways and Means (1998, 867) and Joint Committee on
Taxation.
TABLE 2

Total EITC Recipients with Children by Range (1994 Tax Year)

 Number of
 Recipients Percentage of
 (In Millions) Recipients
 (2) (3)

Families with children 14.9 78
Phase-in range 4.0 26.6
Maximum credit range 2.1 13.9
Phase-out range 8.9 59.5

 Percentage of PSID
 Families Eligible for
 EITC Benefits in 1992
 (4)

Families with children
Phase-in range 30.94
Maximum credit range 15.26
Phase-out range 53.80

Source: GAO/GGD-96-122BR Earned Income Credit, calculations from p. 20
and 34. Column (4) is calculated using PSID data for families who are
eligible for PSID benefits in 1992. There are 9148 families in the PSID
data set and 1250 families that are eligible for EITC benefits in 1992.
TABLE 3

Distribution of EITC spells

 Families Beginning a Spell Families at a
 Point in Time

Spell Completed Spell Completed Spell
Length Distribution Exit Probability Distribution
in Years D(1) p(t) F(t)
(1) (2) (3) (4)

1 0.50564 0.50564 0.23688
2 0.23613 0.47765 0.22124
3 0.11023 0.42688 0.15492
4 0.05636 0.38079 0.10560
5 0.03224 0.35184 0.07552
6 0.02566 0.43198 0.07212
7 0.01215 0.36021 0.03985
8 0.01011 0.46831 0.03789
9 0.00510 0.44460 0.02151
10 0.00175 0.27525 0.00822
11 0.00113 0.24390 0.00581
12 0.00179 0.51152 0.01004
13 0.00171 1.00000 0.01039

Totals 1.00000 1.00000

Average 2.13462 3.55426

 Families at a Point
 in Time

Spell Uncompleted Spell
Length Distribution
in Years G(t)
(1) (5)

1 0.46847
2 0.23159
3 0.12097
4 0.06933
5 0.04293
6 0.02783
7 0.01581
8 0.01011
9 0.00538
10 0.00299
11 0.00216
12 0.00164
13 0.00080

Totals 1.00000

Average 2.27713

Notes: EITC parameters from Table 1 were used to calculate EITC
benefits. Values were weighted using PSID weights.
TABLE 4

Distribution of EITC Spells (1992 Parameters and Completed Spells Only)

 Families Beginning a Spell Persons at a
 point in Time

Spell Length Completed Spell Completed Spell
in Years Distribution Exit Probability Distribution
(1) (2) (3) (4)

 1 0.53469 0.53469 0.24092
 2 0.20639 0.44356 0.18600
 3 0.09777 0.37763 0.13217
 4 0.05645 0.35028 0.10173
 5 0.02986 0.28521 0.06727
 6 0.02030 0.27131 0.05489
 7 0.01800 0.33011 0.05678
 8 0.01145 0.31340 0.04127
 9 0.00909 0.36236 0.03686
10 0.00515 0.32195 0.02320
11 0.00353 0.32554 0.01750
12 0.00320 0.43733 0.01730
13 0.00412 1.00000 0.02411

Totals 1.00000 1.00000

Average 2.21932 4.087

 Persons at a point
 in Time

Spell Length Uncompleted Spell
in Years Distribution
(1) (5)

 1 0.45059
 2 0.20966
 3 0.11667
 4 0.07261
 5 0.04718
 6 0.03372
 7 0.02457
 8 0.01646
 9 0.01130
10 0.00721
11 0.00489
12 0.00330
13 0.00185

Totals 1.00000

Average 2.543
TABLE 5

Recidivism for EITC Exits

 Cumulative
 Recidivism Percentage of Percentage of
Years Rate Households Eligible Households Eligible
Since Exit (R(t)) to Return to ELTC to Return to EITC
(1) (2) (3) (3)

1 0.1601 0.16014 0.16014
2 0.3095 0.07023 0.23037
3 0.2951 0.04625 0.27662
4 0.2899 0.03202 0.30863
5 0.2744 0.02152 0.33016
6 0.2590 0.01475 0.34490
7 0.3080 0.01299 0.35789
8 0.3740 0.01091 0.36881
9 0.2053 0.00375 0.37256
10 0.4634 0.00673 0.37929
11 0.1756 0.00137 0.38065
12 0.3810 0.00245 0.38310
13 1.0000 0.00398 0.38708

Total returning 0.387
TABLE 6

Recidivism for EITC Exits (1992 Parameters and Completed Spells Only)

 Cumulative
 Pcrcentage of Percentage of
Years Receidivism Households Eligible Households Eligible
Since Exit Rate to Return to EITC to Return to EITC

 1 0.1746 0.17459 0.17459
 2 0.3770 0.08092 0.25552
 3 0.3155 0.04220 0.29772
 4 0.3279 0.03002 0.32774
 5 0.3507 0.02158 0.34932
 6 0.2806 0.01121 0.36053
 7 0.3127 0.00899 0.36951
 8 0.3908 0.00772 0.37723
 9 0.2864 0.00345 0.38068
10 0.5290 0.00454 0.38522
11 0.2218 0.00090 0.38612
12 0.1086 0.00034 0.38646
13 1.0000 0.00281 0.38927

Total returning 0.389
TABLE 7

Distributions of EITC Spells in the Phase-in Range and Phase-out/Maximum
Credit Ranges (1992 Parameters)

 Phase-in Range Phase-out and
 Maximum Credit
 Ranges

 Families Families
 Beginning Families at a Beginning
 a Spell Point in Time a Spell

Spell Completed Completed Completed
Length Spell Exit Spell Spell
in Years Distribution Probability Distribution Distribution
(1) (2) (3) (4) (5)

 1 0.66565 0.66565 0.42857 0.54931
 2 0.21463 0.64192 0.27637 0.22558
 3 0.07369 0.61549 0.14233 0.09706
 4 0.02239 0.48643 0.05767 0.04513
 5 0.01194 0.50487 0.03843 0.02349
 6 0.00539 0.46067 0.02083 0.02060
 7 0.00122 0.19271 0.00548 0.01025
 8 0.00194 0.38065 0.00999 0.01104
 9 0.00056 0.17708 0.00324 0.00559
10 0.00204 0.78481 0.01313 0.00441
11 0.00056 1.00000 0.00396 0.00246
12 0.00165
13 0.00344

Totals 1.00000 1.00000 1.00000

Average 1.55319 2.278 2.05922

 Phase-out and Maximum Credit
 Ranges

 Families
 Beginning Families at a
 a Spell Point in Time

Spell
Length Exit Completed
in Years Probability Spell
(1) (6) (7)

 1 0.54931 0.26676
 2 0.50053 0.21909
 3 0.43116 0.14140
 4 0.35241 0.08766
 5 0.28323 0.05703
 6 0.34654 0.06002
 7 0.26395 0.03485
 8 0.38605 0.04288
 9 0.31843 0.02443
10 0.36877 0.02142
11 0.32640 0.01317
12 0.32407 0.00961
13 1.00000 0.2171

Totals 1.00000

Average 3.688
TABLE 8

Percentage of Working and Not Working Heads of Household Using 1992
Rules (Percentages are for the Year the Family Begins a Spell)

 All Phase-in Phase-out and Max
 Qualifiers Range Credit Ranges
 (2) (3) (4)

1. Not working 28.72 49.24 17.7
2. Working 71.28 50.76 82.3
3. Average hours worked 1536 920 1782
TABLE 9

Percentage Distribution for Beginning Types of all Spells of EITC (1992
Rules)

 Percentage of all Beginnings

 Phase-out and Max
 All Qualifiers Phase-in Credit Ranges
Beginning Type (2) (3) (4)

1. Wife is now head 4.63 6.17 6.29
2. Entered sample
 qualifying in 1975 11.63 5.78 10.06
3. Family now has
 eligible child 11.83 6.87 9.72
4. Lower earnings 52.62 57.31 46.59
5. Higher earnings 18.89 23.49 26.80
6. All other 0.40 0.39 0.50

Total 100.00 100.00 100.00
TABLE 10

Percentage Distribution for Why Households Become Ineligible to Receive
EITC (1992 Rules)

 Percentage of all Endings


 All Qualifiers Phase-in
Ending Type (2) (3)

1. Head married or wife remarried 7.36 5.15
2. Exited sample qualifying in 1992 18.79 15.75
3. Last eligible child gone 15.30 12.49
4. Higher earnings 43.08 46.80
5. Lower earnings 15.38 19.72
6. All other 0.10 0.09

Total 100.00 100.00

 Percentage of all
 Endings

 Phase-out and Max
 Credit Ranges
Ending Type (4)

1. Head married or wife remarried 7.95
2. Exited sample qualifying in 1992 15.94
3. Last eligible child gone 11.28
4. Higher earnings 43.86
5. Lower earnings 20.71
6. All other 0.25

Total 100.00


(1.) See Hoffman and Seidman (1990) and Eissa and Liebman (1996).

(2.) Before 1994, to receive the EITC, families had to have an eligible child. However, in 1994 the EITC was expanded to include households without children. However, the maximum benefit in 1996 was only $323 a year.

(3.) See Killingsworth (1983).

(4.) For example, Robert Barro (1996, A18) wrote: "It makes sense that a comprehensive welfare program would include EITC-type payments to the working poor. A subsidy rate in excess of 40% at the low end ought also be considered, although this change would require more budget outlays and create a greater disincentive to work in the phase-out range." Gary Becker (1996, 22) wrote: "This program (EITC) rewards rather than penalizes poor families with working members." Richard Burkhauser (1996, 19:1) wrote: "Thus, the EITC ensures that minimum-wage workers who live in poor families will, in fact, receive net wage income well above the current minimum wage of $4.25 per hour."

(5.) Exceptions may include Eissa and Liebman (1996) and Meyer and Rosenbaum (2000), who argue that the EITC may give single mothers an incentive to start working.

(6.) For example, see Hoffman and Seidman (1990), Eissa and Liebman (1996), and Liebman (1998).

(7.) The percentage of recipients in the phase-in range is from calculated from General Accounting Office (GAO) (1996).

(8.) The 1998-2001 parameters were supplied by Al Davis, a staff member for the Committee on Ways and Means, U.S. House of Representatives.

(9.) This assumes that the family stays in the phase-in range.

(10.) $953 equals 0.14 times $6,810.

(11.) See Hoffman and Seidman (1990) for a discussion on how the EITC is calculated.

(12.) The only exceptions were families who worked full-time with two or more children in 1995 and 1996.

(13.) See Liebman (1998) and Committee on Ways and Means (1998).

(14.) According to the GAO (1996), in tax year 1994, $21.2 billion was spent on the EITC and there were 19.1 million recipients. Between tax years 1990 and 1994, EITC recipients increased by approximately 50%, from 12.6 million to 19.1 million, and in real terms expenditures increased by 150% (GAO, 1996, 15).

(15.) The EITC started in 1975.

(16.) The amount of any potential EITC benefits were not used to calculate income.

(17.) There were 1250 PSID families who were eligible to receive the EITC in 1992 and 9148 families for an unweighted average of 13.7%.

(18.) In 1994, there were 19.1 million EITC recipients. Twenty-two percent of the 19.1 million EITC recipients did not have children. However, families with no children received only 3% of the total EITC credits. Their average credit was $168. On the other hand, 43% of EITC recipients had one qualifying child and received 49% of all credits, and 36% of ELTC recipients had two or more qualifying children and received 48% of all credits. This study will only include families with at least one qualifying child.

(19.) These calculations of EITC spells are based on Bane and Ellwood's (1983, 1994) calculations of AFDC spells.

(20.) 24.05% of all households were eligible to receive EITC benefits for at least one year sometime during the 18-year period.

(21.) If after a year of not being eligible to receive the EITC the family becomes eligible again, they begin a new spell.

(22.) Likewise, in the case of unemployment. Most people are unemployed for a short period of time. However, a few people will be unemployed for a long period of time. Even though the long-term unemployed are a small fraction of those people who become unemployed, because they are unemployed longer they end up being a larger fraction of the unemployed.

(23.) In this example, a spell in the hospital would be the number of days that a person is in the hospital, whereas in the case of the EITC, eligibility is calculated in years.

(24.) For a longer discussion of equations (1), (2), and (3) see Bane and Ellwood (1983).

(25.) The distribution of completed spells for those eligible to receive the EITC at a point in time is the distribution of new spells weighted by duration and renormalized.

(26.) Whether the results were calculated with uncompleted spells but with 1992 parameters or with the actual parameters for each year with completed spells only, the results were similar to Table 4.

(27.) Real negative income taxes may have an earnings disregard at the bottom, so the maximum credit range and the phase-out range together may be thought of as a negative income tax.

(28.) Of course, a family with two earners could have a head who was not working but the spouse was. On the other hand, both the spouse and the head may not be working at the time of the interview.

(29.) The 1992 parameters are used to calculate Tables 9 and 10. However, when annual data was used to calculate Tables 9 and 10, eligibility rule changes did not seem to have a major effect on households entering or leaving the EITC.

REFERENCES

Barro, R. J. "Workfare Still Beats Welfare." Wall Street Journal, 21 May 1996, A18.

Bane, M. J., and D. T. Ellwood. "The Dynamies of Dependence: The Routes to Self-Sufficiency." Report to the U.S. Department of Health an Human Services, Urban Systems Research and Engineering, Cambridge, MA, 1983.

_____. Welfare Realities: From Rhetoric to Reform. Cambridge, MA: Harvard University Press, 1994.

Becker, G. S. "How to End Welfare As We Know It'-- Fast." Business Week, 3 June 1996, 22.

Burkhauser, R. V. "Move Over, Minimum Wage; Tax Credits Are on Target." Christian Science Monitor, 14 February 1996, 19:1.

Committee on Ways and Means, U.S. House of Representatives. Overview of Entitlement Programs. Y4.W36:l0-4/993. Washington, DC: Government Printing Office, 1993.

_____. Overview of Entitlement Programs. Y4.W36:104/998. Washington, DC: Government Printing Office, 1998.

Eissa, N., and J. B. Liebman. "Labor Supply Response to the Earned Income Tax Credit." Quarterly Journal of Economics, 111(2), 1996, 605-37.

General Accounting Office. "Earned Income Credit: Profile of Tax Year 1994 Credit Recipients." Report to the Chairman, Committee on Finance, U.S. Senate, GAO/GGD-96-122BR, June 1996.

_____. "Earned Income Credit: Claimants' Credit Participation and Income Patterns, Tax Years 1990 through 1994." Report to the Chairman, Committee on Finance, U.S. Senate, GAO/GGD-97-69, May 1997.

Hoffman, S. D., and L. S. Seidman. The Earned Income Tax Credit: Antipoverty Effectiveness and Labor Market Effects. W E. Upjohn Institute for Employment Research, Kalamazoo, MI, 1990.

Killingsworth, M. Labor Supply. Cambridge: Cambridge University Press, 1983.

Liebman, J. B. "The Impact of the Earned Income Tax Credit on Incentives and Income Distribution," in Tax Policy end the Economy, vol. 12, edited by J. M. Poterba. Cambridge, MA: MIT Press, 1998, pp. 83-120.

Meyer, B. D., and D. T Rosenbaum. "Making Single Mother Work: Recent Tax and Welfare Policy and Its Effects." National Tax Journal, part 2, 53(4), 2000, 1027-61.

Panel Study of Income Dynamics. Interuniversity Consortium for Political and Social Research, Ann Arbor, MI, 1992.

ABBREVIATIONS

AFDC: Aid to Families with Dependent Children

CPI: Consumer Price Index

EITC: Earned Income Tax Credit

GAO: General Accounting Office

PSID: Panel Study of Income Dynamics

John B. Horowitz *

* Special thanks to Dr. Edgar Browning, Dr. Eric Helland, Dr. Cecil Bohanon, and Dr. Gary Santoni and to Tom Charles for his able research assistance. This research was supported by a grant from the John W. Fisher Faculty Fellowship at Ball State University.

Horowitz: Associate Professor, Department of Economics, Ball State University, Muncie, IN 47306. Phone 1-765-285-1436, Fax 1-765-285-8024, E-mail jhorowitz@bsu.edu
COPYRIGHT 2002 Western Economic Association International
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2002 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Horowitz, John B.
Publication:Economic Inquiry
Geographic Code:1USA
Date:Jul 1, 2002
Words:9178
Previous Article:Do minimum wages fight poverty?
Next Article:Provider selection, bargaining, and utilization management in managed care.
Topics:

Terms of use | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters