Full text of Working Papers (Federal Reserve Bank of Chicago) : The Changing Pattern of Wage Growth for Low Skilled Workers, Working Paper 2005-24

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

Federal Reserve Bank of Chicago

The Changing Pattern of Wage
Growth for Low Skilled Workers
Eric French, Bhashkar Mazumder and
Christopher Taber

REVISED May 3, 2006
WP 2005-24

The Changing Pattern of Wage Growth for Low Skilled Workers

Eric French
Federal Reserve Bank of Chicago
Bhashkar Mazumder
Federal Reserve Bank of Chicago
Christopher Taber
Department of Economics
and
Institute for Policy Research
Northwestern University1

May 3, 2006

1.

Introduction
One of the fundamental facts in labor economics is that, on average, wages tend to rise rapidly early

in a worker’s career. Since wage growth during the early stages of the one’s career provides a potential
pathway out of poverty, it is important to understand what causes this wage progression and how it is
affected by changes in the overall economy. In this chapter, we focus on the key components that
determine an individual's early career wage growth and how these factors have changed for less skilled
workers over the last twenty years. In particular, we examine the relative importance of accumulating
work experience as compared to the quality of job matches in influencing wage growth over this time
period.
The importance of experience accumulation on wage growth is a particularly relevant concern for
policy makers in light of the reforms to the tax and welfare system that have taken place since the early
1990s. The expansion of the Earned Income Tax Credit (EITC) and the Personal Responsibility and
Work Opportunity Act (PROWORA), were specifically designed to encourage low skilled individuals to
enter the workforce. Perhaps as a result of these policies, the labor force participation rates of men and
women with low levels of schooling did in fact increase during the 1990s. However, entry-level wages
for low-skilled workers are low and have been stagnant over the last twenty five years.2 This calls into
question whether these programs can really do much to alleviate poverty.
Although starting wages of low skill workers are low, they potentially increase with experience.
Therefore, the success of tax and welfare reforms in reducing poverty rests critically upon the extent to
which experience accumulation increases the wages of low skilled individuals. To highlight the
importance of experience accumulation, consider Figure 1, which shows the extent to which wages
increase with age.3 Between ages 18 and 28 wages grow by about 45 percent, from $6.90 to $10 per hour,
for men with no college. For a single earner in a 4 person household working 40 hours per week, 50
weeks per year, such a wage gain would actually move the family out of poverty. Clearly, labor force
experience has the potential to substantially raise wages.

2

However, labor market experience is not the only potentially important source of wage growth for
younger workers. Robert Topel and Michael Ward (1992) show that for workers who entered the labor
market in the late 1950s, earnings gains at job switches accounted for about a third of early career
earnings growth. Arguably, the earnings increases associated with job switches reflect improvements in
the quality of job matches over an individual’s career. Concerns about job match quality might be
especially important for low skill workers, because they have less stable employment patterns. As a
result, some observers have expressed concern that difficulties in holding jobs, and in moving to better
ones (Harry Holzer and Robert LaLonde, 1999), retards wage growth for low skill workers.
One would expect the job match process to change over the business cycle in ways that would lead
this component of wage growth to be highly cyclical. First, it is likely to be much easier to find a good
job during a boom than during a recession. Second, since layoffs are typically associated with wage
declines one would expect the higher rate of layoffs during recessions to depress wage growth. This
suggests that an analysis of the changing pattern of wage growth should disentangle the importance of
finding a good job, or matching, from the importance of work experience in determining wage growth.
Despite the centrality of early career wage growth to labor economics, it is surprising how little we
actually know about the factors that determine wage growth, their cyclical properties, and how they have
changed over time.4 The primary purpose of this chapter is to document how wage growth has changed
over the last twenty years for low skilled workers and to understand which components have driven these
changes. To motivate our analysis we present some basic facts concerning early career wage aged growth
for low skilled workers. Figure 2 plots hourly wage growth for workers aged 18-28 who did not attend
college using the matched Current Population Survey (CPS) Outgoing Rotation Group data.5 This group,
which contains both men and women, acts as our primary reference group throughout our analysis. The
data shows that young workers’ wages increase about 4.5 percent per year, on average. Figure 2 also
makes clear that wage growth varies considerably over time and may be related to the business cycle.
Wage growth was about 3 percent per year during 1980-1983, 1990-1994, and 2002-2004, the years that

3

unemployment rates were high and rising. In contrast, wage growth averaged 6 percent during 1984-1987
and 1995-2001, the years that unemployment was low and falling.6
For our main analysis we use data from the Survey of Income and Program Participation (SIPP)
covering the years 1984 to 2003. The key advantage of the SIPP is that it allows us to track the same
individuals over several years. This allows us to estimate the rate of job transitions, as well as wage gains
associated with those transitions.
Our empirical strategy is as follows. We first assume that wage changes for continuously employed
workers reflect returns to experience. For workers who switch jobs, wage changes result from both
experience accumulation and changes in the quality of the job match. Using these assumptions we
estimate a baseline model that decomposes wage growth over this period into changes that arise from:
experience accumulation; the returns to experience; the rate of job matching; and the returns to match
quality. As our second specification, we augment the baseline model to also include a common time
effect that affects all individuals’ wages, regardless of experience or match quality. We then decompose
the residual wage growth (net of this common component) into the key factors of interest. We identify the
common effect using wage levels for new labor market entries. The results from both decompositions
allow us to better understand the extent to which the temporal pattern of wage growth is driven by
changes in experience accumulation and matching.
Our main finding that is that wage growth has varied considerably over the last 20 years. We find
that the vast majority of the variation in wage growth is due to variability in the return to experience over
time.7 Although the return to experience seems to change from year to year, there is no strong evidence of
a secular trend. On average over this time period, an additional year of experience increases wages about
4 percent but this gain varies from as much as 6 percent to as little as 2 percent. In contrast, essentially
none of the changing pattern of wage growth can be attributed to changes in experience accumulation, job
to job finding rates, or layoff rates. While these variables have the expected cyclical relationships, the
overall magnitude of their contributions to changes in wage growth is extremely small. Therefore, we

4

find that the return to experience is much more important than job matching for explaining the variability
in early career wage growth. We know of no other work that has examined this question.
The extent to which the wage growth pattern varies with the business cycle depends crucially on our
specification. In our base model (which does not account for common time effects), we find that the
return to experience is strongly procyclical. In our second model (that does account for common time
effects), we find that the cyclicality of returns to experience results fully from the wage level effects. That
is, after we account for the common time effect, we find no remaining relationship between the returns to
experience and the business cycle. We are unaware of any previous work that documents how the
determinants of wage growth vary over the business cycle, let alone for any particular subgroup of the
population.
We also estimate the model separately by skill groups and gender. While our estimates by subgroup
are noisy, we find that average returns to experience and matching, and their cyclical and secular patterns,
do not vary systematically by education or gender. Thus, we see no evidence that increasing differences
in the return to experience and job matching across skill groups plays an important role in explaining
rising wage inequality.
Our results suggest that policy makers should not be overly concerned that workers who enter the
labor market during a recession will receive less wage growth over their first ten years than workers who
enter during a boom. This suggests that the recent emphasis of policymakers to promote working among
young low skilled workers is not misplaced. Furthermore, we find that for an average worker, experience
accumulation is much more important than job matching in determining early career wage growth.
Nonetheless, policies that are able to deliver especially high quality job matches may also be desirable.
Section 2 discusses the relationship between our results and previous work while Section 3 describes
our methodological approach. We describe the data in Section 4 and the empirical results in Section 5. In
Section 6 we summarize our main results and provide some conclusions.

2.

Related Work

5

Although we are unaware of any previous study that has analyzed the changing pattern of lifecycle
wage growth and its determinants, our research does intersect with a wide range of studies on related
topics. There is a long literature in labor economics that has studied the age-earnings profile beginning
with Jacob Mincer (1962) and Yoram Ben Porath (1967). These studies have shown that earnings
increase rapidly early in the lifecycle (i.e., between ages 18 and 28) as workers invest in their human
capital by gaining experience on the job.8 In other words, they are more productive because they have
more skills, and they are compensated accordingly.
A second literature has emphasized the importance of job matching in explaining lifecycle wage
growth. Workers are able to find better matches for their skills over time. For example, since searching
for a job is time consuming, an individual straight out of school may not wait until she finds the best
match for her skills, but instead takes a job she can find quickly. Over time, however, she may find better
matches for her skills. A substantial literature has arisen that examines the effects of job mobility (or job
stability) on earnings using either regression analysis or structural modeling. Examples include Jacob
Mincer and Boyan Jovanovic (1981), Christopher Flinn (1986), John Antel (1991), Pamela Loprest
(1992), Robert Topel and Michael Ward (1992), Kenneth Wolpin (1992), Farber (1994), and Jacob
Klerman and Lynn Karoly (1994). Theresa Devine and Nicholas Kiefer (1991) and Kenneth Wolpin
(1995) provide surveys of different aspects of these literatures. These papers show that turnover is an
important engine for wage growth.
Another strand of the literature has focused specifically on wage growth among low wage workers.
Fredrik Andersson, Harry Holzer, and Julia Lane (2005) make use of a unique matched worker/firm data
set to study job mobility for this particular group. They provide a very detailed analysis of worker
mobility and wage progression demonstrating, among other things, the importance of turnover as a
component of wage growth.
A number of papers have studied the relationship between the determinants of lifecycle wage growth
and skill levels. Tricia Gladden and Christopher Taber (2000, 2004) focus on younger workers and
compare the wage growth of medium skill workers with wage growth of low skill workers and find that

6

they are similar. Helen Connolly and Peter Gottschalk (2000) look across all age groups of workers and
find that more educated people receive higher returns to both tenure and experience. Since these papers
use somewhat different concepts and different comparison groups they are not directly comparable.
Another related literature has examined how labor market transitions have changed over the business
cycle. In a well known study, Steven Davis, John Haltiwanger, and Scott Schuh (1996) use plant level
data and find that job destruction is strongly countercyclical in the manufacturing sector. In contrast, Eva
Nagypal (2004) and Robert Shimer (2005) utilize individual level data on workers in all sectors of the
economy and find that job separation rates are not countercyclical.9 These authors also document that job
finding rates and job to job transitions are highly pro-cyclical.
There is an extremely large literature on the cyclicality of aggregate wage levels. Katherine Abraham
and John Haltiwanger (1995) provide a nice summary of the literature concluding that aggregate wages
are slightly pro-cyclical, but have become more so in recent years. They also suggest that wages are procyclical at the individual level, but that aggregate wages can mask this fact due to composition effects.
This is because low wage workers, whose employment patterns are more pro-cyclical, constitute a larger
fraction of the labor force in booms than in busts. This selection effect lowers average wages in booms.
A few papers have studied the effects of economic conditions early in one's career on earnings later in
life. Paul Beaudry and John DiNardo (1991) are not interested in wage growth per se, but rather in
contracting over the business cycle. They show that the lowest unemployment rate during an individual's
tenure at a firm is a strong predictor of current wages. David Neumark (2002) examines the effects of job
turnover on future wages. In doing so, he uses local labor market conditions when an individual is young
as an instrument for turnover. He finds that job stability at young ages leads to higher earnings at older
ages. Phil Oreopolous, Till von Wachter and Andrew Heisz (2006) find that there are large initial
earnings losses from graduating college during a recession that dissipate after about 8 to 10 years. They
find evidence of heterogeneous affects, for example, relatively lower skilled college graduates suffer
larger and more permanent earnings losses.

7

3.

Methodology
In this section we present our framework for modeling wage growth. Since previous research has

shown that experience accumulation and job transitions are associated with wage growth, we model both
labor market transitions and wage dynamics. We assume that time is discrete and denoted by t, which in
practice will be monthly. Let jit denote the job held by individual i at time t. To keep the notation simple
we denote non-employment with a zero; so that jit = 0 means that individual i was not working at time t.
Therefore, for individuals who are working at time t – 1 (i.e. jit-1 ≠ 0), there are three possible labor market
transitions: stay on the same job, become non-employed, or switch to a new job. We write these three
transition probabilities respectively as
Pr(jit = jit-1| jit-1 ≠ 0)

(1)

Pr(jit = 0 | jit-1 ≠ 0)
Pr(jit ≠ jit-1, jit ≠ 0 | jit-1 ≠ 0).
Clearly these must sum to one.
Individuals who are not employed at time t - 1 start a new job with probability Pr(jit ≠ 0 | jit-1 = 0) and
fail to start a new job with probability Pr(jit = 0 | jit-1 = 0). Again, these two probabilities must sum to one.
Our primary focus is the relationship between labor market transitions and wage dynamics. Let

ωit

denote the wage of individual i at time t. A key feature of our data (which will be discussed in the data
section) is that we do not use wages from every month. As a result, the number of months between wage
observations will vary across observations. With this in mind, define the l th period difference operator
as ∆l , e.g. ∆lωit = ωit − ωit − l . We consider two different empirical specifications for wage changes. In
our first approach we assume that the structural (i.e., true) model10 of wages is:

∆lωit = β t∆lΑit + ∆lηijit + ∆lεit ,

(2)

where Αit represents actual experience (and thus ∆lΑit is the amount of accumulated work experience
between time periods t - l and t), ηijit is a match specific component between individual i and job jit , and

εit represents an error term which is orthogonal to the other components of the model. We assume that
8

the quality of a match does not change unless an individual changes jobs so that for job stayers ∆lηijit = 0
by assumption. We are also ignoring any effect from job tenure so that ∆lηijit may also include any
effects from the loss of job tenure. Note that we assume that wage growth is linear in experience which is
a reasonable approximation given that we only analyze young workers.
In our second specification we allow for a common wage level effect which affects all
individuals' wages regardless of their experience or the quality of their job match. To motivate this
specification suppose that wages are determined by the pricing equation

ω = Rt H it
it

where Rt is the rental rate of human capital at time t, and H it is the amount of human capital for
individual i at time t. Taking logs we define αt = log(Rt), which we refer to as the common time effect .
We assume that ∆l log( Hit ) = γt∆lΑit + ∆lηijit + ∆lεit , and re-define ω it to be the log wage so that

∆lωit = ∆lαt + γt∆lΑit + ∆lηijit + ∆lεit .

(3)

In this framework, fluctuations in wage changes may be due to one of several factors. First, it may be that

∆lαt changes over time, affecting everyone’s wage, regardless of human capital level. For example, in
Robert Hall’s model (2005, this volume), ∆lαt represents the growth rate of technological improvements
or growth in the amount of capital used by each worker. Macroeconomists typically focus on αt as the
sole factor in explaining wage fluctuations (see Katherine Abraham, and John Haltiwanger, for example).
This is a reasonable first approximation since for much of US history wage gains were shared by all skill
groups. However, as Robert Hall points out, his model fails to account for many recent changes in the
distribution of wages such as the well documented increase in wage dispersion. Moreover, as we pointed
out in Section 2, there is ample evidence that other factors influence wage growth as well. These other
factors are the focus of the paper.
A second explanation for shifts in wage growth is that the return to experience may change over time
(that is to say, through β t in equation (2) or γt in equation (3)). This might be the case if the rate at

9

which individuals learn on the job varies over time, for example.11 A third possibility is that changes in
the amount of experience accumulation (that is to say, ∆lΑi ) may result in changing wage growth. We
generally would expect this component to be procylical because during recessions, individuals who are
unemployed are not gaining work experience. Thus, one would expect this effect to lead overall wage
growth to be higher during a boom.
A fourth factor in explaining variation in wage growth is changes in the quality of job matches over
time (that is to say, Ε(∆lηijit ) ). We examine this empirically by looking at both job switchers that go
directly from one job to another as well as job switchers who experience an intervening spell of
unemployment. Direct job switches lead to changes in wage growth either because the rate of job
switching changes or because the wage gains associated with each switch changes. As Robert Shimer
(2004) and Eva Nagypal (2004) point out, job-to-job switching is pro-cyclical. Given that wages tend to
rise with job-to-job transitions, increased job-to-job transitions will raise wages even if the average wage
gain associated with these job changes does not change over time. Of course, it could also be the case
that the average wage gain associated with job switching (the average change in the value of ηijit given a
job change) has changed over time. Similarly job changes involving an unemployment spell are likely to
lead to negative wage growth either because those transitions (which typically involve wage losses) are
countercyclical or because the amount of the wage losses are countercyclical.
When confronted with estimating equation (2) or (3), several identification issues arise. First, (as with
all wage regressions) there is a selection issue because wage growth is only observed for workers who are
employed in both periods. We assume that sample selection is based on an individual fixed effect, which
is differenced out, but that there is no selection on ∆lε it . That is, we allow individuals with permanently
low productivity to have different participation rates than those with permanently high productivity.
However, we do not allow short term wage fluctuations from changes in εit to affect the decision to
work.

10

A second problem lies in estimating the returns to experience ( β t in equation (2) or γt in equation
(3)). For many reasons we would expect job matches (ηijit ) to be correlated with experience ( Αit ). For
example, a worker who has been unemployed for an extended period of time may become less choosy
about jobs, and will accept a worse match. This is problematic because it is very difficult to measure
match quality.12 As a result, running a regression of wage growth on the change in experience without
including match quality will not yield a consistent estimate of the returns to experience. Instead we use an
alternative approach where we utilize only the sample of workers who do not switch jobs (stayers) so that

∆lηijit = 0 . Under model (2), one can write
Ε(

∆lωit
βt∆lΑit + ∆lεit
| ∆ljit = 0) = Ε(
| ∆ljit = 0)
∆lΑit
∆lΑit

(4)

= βt

(∆ε

where Ε

l it

∆lΑit

| ∆ljit = 0) = 0 by assumption. Thus, in principle, we can estimate βt consistently by

simply taking the sample mean of

∆lωit
for stayers in each month. Since the sample sizes are too small
∆lΑit

to estimate this parameter month by month, we smooth across months using kernel regression.
Separately identifying the change in the common time effect ( ∆lαt ) from the return to experience
( γt ) in model (3) is not straight forward. The problem is that for continuously employed workers, if we
measure wage changes over a fixed time period, say 4 months, there is no variation across this group in
the amount of experience accumulated ( ∆lΑit ). As a result it is impossible to separately identify these
two components using wage changes on stayers as we did in equation (4).13 One approach would be to
use movers with spells of non-employment to try to separately identify these parameters. However, as we
pointed out previously the change in experience is likely to be correlated with the change in match quality
which we are unable to measure very well. Therefore, in order to obtain consistent estimates of the
parameters of the model, we take an alternative approach. First, we estimate αt , the common time
component to wages, using the wages of new labor market entrants because they all have zero experience.

11

Assuming that the quality of new workers, θi (where θi is discussed more formally in endnote 8), and the
quality of new matches does not change over time, changes in the wages of new entrants change only
because of changes in αt . That is we assume that we can write expected wages of new entrants as

Ε[ωit | Αit = 0] = αt + Ε[θi + ηijit + εit | Αit = 0]
= αt + a constant.
Because wages of new entrants are equal to αt (plus a constant that can be differenced away), changes in
average wages of entrants yield consistent estimates of ∆ l α t . We use monthly data to estimate αt and
smooth as above with kernel regression. For this estimation procedure we use the larger CPS sample
rather than SIPP in order to obtain more precise estimates.
In the second stage, now that we have estimated αt , we use a strategy for estimating γt that is
analogous to (3).

Ε(

ˆ
γt∆lΑit + ∆lεit
∆lωit − ∆lαt
| ∆ljit = 0) = Ε(
| ∆ljit = 0)
∆lΑit
∆lΑit

(5)

= γt .
Here we take the sample mean of

ˆ
∆lωit − ∆lαt
smoothing with kernel regression. Compared to our
∆lΑit

estimation of the first model that does not account for a possible common time effect, we have simply
taken our estimates of wage growth and first subtracted out our estimates of changes in the common
component of wages before applying our statistical procedure.
We should point out some problems that arise with our approach to estimating αt . The fundamental
problem is that the value of the unobserved components of workers’ wages ( θi + ηijit + εit ) may change
across new entrant cohorts for one of several reasons. For example, it could be that the average individual
fixed effect ( θi ) changes from cohort to cohort, because of changes in the quality of education. A second,
and likely more serious issue, is sample selection bias. To be in included in our sample, an 18 year old
must be working and not in school. However, there have been both secular and cyclical changes in labor

12

force participation rates and college attendance rates of young individuals. Most notably, labor force
participation rates for 18 year olds have been falling over our sample period.
Since we do not believe that it is possible to perfectly separate changes in the common aggregate
component from the returns to experience, we present results using models both with and without our
estimates of αt and allow the reader to choose their preferred specification. One can interpret the
estimates without αt as incorporating both time and experience effects. The results with αt try to
separate these two processes, although this separation is not as clean as one might like. We show below
that many of our main results are qualitatively unaffected by the assumptions used to obtain ∆lαt .
However, we find one very important difference which is that experience growth is strongly procyclical in
the first specification, but this procyclicality disappears when we control for ∆lαt .
Our next goal is to summarize the importance of changes in job matches on wage growth. Since it is
very difficult to measure changes in match quality ( ∆lηijit ), as we pointed out earlier, we simply estimate
the expected wage gains for two types of switchers: job to job switchers and job-to-nonemployment to job
switchers. For brevity, we describe our methodological approach for our second model where we do not
include αt . Define Νitl to be an indicator of whether person i experienced a nonemployment spell
between periods t − l and t . For a job to nonemployment to job switcher, the average wage gain at
switching is

Ε(∆lηitit | jit ≠ jit − l, Νitl = 1) = Ε(∆lωit − γˆt∆lΑit | jit ≠ jit − l, Νitl = 1) ,

(6)

where the expectation is over all possible durations of non-employment. We estimate this just by taking

ˆ
the sample mean of ( ∆lωit − γt∆lΑit ) for job to non-employment to job switchers in each month (using
kernels to smooth). Basically, we take wage growth in each period for this sample and subtract out our
estimates of the change in the common time component (estimated earlier) and also subtract out the
change in wages due to experience. Similarly we can identify the average wage gain at switching for job
to job switchers using

Ε(∆lηijit | jit ≠ jit − l , Νitl = 0) = Ε(∆lωit − γˆt∆lΑit | jit ≠ jit − l, Νitl = 0) . (7)
13

In words this is the average wage gain that occurs at job to job switches. We estimate this in a manner
analogous to the job to non-employment to job switchers.
Given our estimates of the various sources of wage growth, our next goal is to decompose overall
wage growth into its various components using Oaxaca style decomposition. We describe the
decomposition for the first model in which we do not incorporate αt . The extension to the model
including αt is straightforward in that we just perform the same decomposition of wage growth net of
changes in aggregate wage levels (i.e. we decompose E (∆1 log(H it )) = E (∆1ωit − ∆1αt ) rather than

Ε[∆1ωit ] ). One issue that arises in this type of decomposition is the definition of wage growth among
workers who are not working. To keep the model as simple as possible, for a worker who is not working,
we define their implicit wage (or match component) as their wage (or match component) on their previous
job. Thus wage growth is 0 by definition for a non-employed worker, but is nonzero when they start their
new job. We also note that wage growth at time t is well defined only for individuals who worked at some
point prior to t. We leave this conditioning implicit as every expectation we write below conditions on
individuals who worked at some point prior to time t. Under this normalization we can write

Ε(∆1ωit ) = β tΕ(∆1Αit ) + Ε(∆1ηijit | jit ≠ 0, jit − 1 = 0) Pr( jit ≠ 0, jit − 1 = 0)

+ Ε(∆1ηijit | jit ≠ jit − 1, jit − 1 ≠ 0) Pr( jit ≠ jit − 1, jit − 1 ≠ 0)
The first component is the wage growth due to experience gained on the job, the second represents the
change (likely negative) associated with job to non-employment to job changes, and the third represents
the wage gains that occur at job to job transitions. Since our SIPP panels are relatively short (2 to 4 years)
and since durations of non-employment can sometimes be very long, we estimate the three transition rates
described in equation (1) and use these to simulate the probability of a job to non-employment to job
transition.14
It is easiest to explain the decomposition this if we stack the parameters and write this in vector
notation:

14

βt
Ε(∆1Αit )




 Ε(∆1ηijit | jit ≠ 0, jit − 1 = 0) , Χt =  Pr( jit ≠ 0, jit − 1 = 0) 
Gt = 



Pr( jit ≠ jit − 1, jit − 1 ≠ 0)
Ε(∆1ηijit | jit ≠ jit − 1, jit − 1 ≠ 0)





(8)

We denote the mean of parameters over years as

Χ≡

1 Τ
∑ Χt .
Τ t =1

The advantage of this additional notation is that it allows us to express the decomposition in the following
way

Ε(∆1ωit ) = G ′tΧt

(9)

= G ′tΧ + G ′t ( Χt − Χ ) .
The first part of this decomposition G ′tΧ reflects how much of the wage change can be explained purely
by changes in the coefficients over time. The remaining component contains the amount left which is due
to labor market transitions changing over time.
We decompose the result even further. In each year the value of G ′tΧ is a sum of three separate
components (associated with change in experience, number of job to job transitions, and number of job to
non-employment to job transitions). We present each of these three components later in the chapter.
We estimate the regressions above and perform the decompositions above allowing the coefficients to
vary over time and across various demographic groups.
We should point out here that this decomposition should be interpreted as descriptive. Separating the
returns to tenure from the returns to experience and dealing with the many sample selection problems
inherent with turnover and labor supply among low wage workers is beyond the scope of this work. Our
Results should be interpreted as a statistical decomposition of the observed wage growth in the data rather
than causal effects.

4.

Data

15

For our main analysis, we use pooled data from the Census Bureau’s Survey of Income and Program
Participation (SIPP). The SIPP surveys are a series of two to four year panels that began in 1984. The
SIPP panels from 1984 through 1993 consist of a national sample of around 20,000 households. The 1996
and 2001 SIPP panels include close to 40,000 households. Combining all of the SIPP panels we have
information on all years from 1984 to 2003 except for 2000. The SIPP interviews households every four
months and collects detailed labor market information on all individuals in the household over the
previous four months. Each interview is referred to as a wave. A key aspect of the of the SIPP for our
analysis is that the survey identifies up to two different employers for each individual in each wave and
these job identifiers are consistent across waves.15 As a result we are able to identify cases where
individuals transition by month from: non-employment into work; work into non-employment; stay in the
same job; or, switch employers.16
The SIPP also provides direct data on hourly wages for workers who are paid by the hour. For
salaried workers, we calculate hourly wages for each employer by dividing monthly earnings by usual
hours per week worked times weeks worked in the month.17 This allows us to calculate wage changes
associated with labor market transitions. Hourly wages are deflated to 2003 dollars using the monthly CPI
price index and we drop individuals whose wages are ever below $1. Since the focus of our analysis is on
wage growth early in the career, we confine our sample to individuals between the ages of 18 and 28 who
are never enrolled in school during the time they are in the SIPP. We divide the sample into 3 educational
groups: those who have not completed high school (drop-outs), high school graduate (but no college), and
some college (or more). Our main sample includes over 900,000 person-month observations. It is worth
noting that the time series patterns of unemployment and annual wage growth in the SIPP closely tracks
that found in the CPS.
The fact that the SIPP collects new data every four months is clearly advantageous compared to
annual surveys that suffer from greater recall bias. On the other hand, it is well known that for many
variables, respondents do not accurately report changes that occur each month. Instead changes tend to be
clumped at the seam between the last month of a previous interview and the first month of the next

16

interview. Because of this “seam bias” we only use one wage observation on each interview wave for
workers who are continuously employed with the same firm. Thus for workers who stay on the same job
we construct our wage change measure as ωit − ωit

−l

where l is four (months). We then use the number

of weeks worked in the interval as our measure of [ Αit − Αit − l] .
When a worker works on two different jobs between interviews, we record two wages for that wave
and use both in constructing our wage differential. So for example, suppose a worker was interviewed in
April (call this t = 4). In July they switched to a new job and then were interviewed again in August. We
would gather two wages from the August interview. The wage during the last month (June) on the old job
and the wage from the current job (in August). We then obtain two wage changes in this wave for the
person. First we would take their wage from the last month working at their old job (June which
corresponds to t = 6) and subtract the wage from the previous wave ω 6 − ω 4 . The person would be
recorded as a stayer in this case because they hadn’t switched employers yet. For the second wage
observation we would use the wage from the interview month (August) and subtract the previous wage
(June) to form ω 8 − ω 6 . This observation would correspond to a job-to-job transition since the person
was working a different job in August than June and was continuously employed. As a result, job
changers are over represented in our sample of wage changes. However, this will not bias our results as
we condition on job changing in the empirical work. Note that given the manner in which the data is
constructed, we will never observe more than one job change in a period in which we obtain wage
differentials.

5.

Results
In the introduction we showed that wages of low skilled young workers grow about 4.5 percent per

year. Using estimates from our two models of wage growth, this section presents evidence on the sources
of that wage growth.

Returns to Experience Accumulation and Job Changes

17

First, we present parameter estimates from equation (2), which does not include the common time
effect αt . The parameters represent the returns to experience (i.e., the average wage change associated
with an extra year of work for stayers), the returns to job to job switches (i.e., the average wage change
associated with a job switch) and the returns to job changes with an intervening non-employment spell.
These are graphed in the top panel of Figure 3a for the combined group consisting of high school
graduates and drop-outs. The returns to experience, β t , depicted with a solid line, vary considerably over
this period but average about 4 percent. The other two lines, representing the wage gains associated with
job to job transitions (dashed line), and job to non-employment to job transitions (vertical lines), also vary
over time. On average, wages rise about 3 percent during job to job switches and decline about 3 percent
during job switches with an intervening non-employment spell.
These results are similar in magnitude to those found by Helen Connolly and Peter Gottschalk (2000)
who include older workers and also use the SIPP pooling data over the 1986-1993 period. For example,
for workers with high school or less they find that wage gains at job to job changes are around 3.5 percent
and that wage losses are just under 3 percent for job to non-employment to job transitions. Our estimates
of the wage losses associated with job to non-employment to job transitions are smaller than Farber’s
(2005) estimates of the wage losses associated with job loss using the Displaced Worker’s Survey, but
show similar time trends. Farber (2005) finds that earnings losses of displaced workers (including
foregone earnings growth from not having a job) declined from 13% in the 1980s to 10% in the early
1990s to 8% in the late 1990s, then increased to 17% in the early 2000s.18
In the bottom panel of Figure 3a we present results that account for changes in the common time
effect on wage levels using the model described in equation (3). The bottom panel of 3a is analogous to
the top except that we also include our estimate of ∆1αt (that is, the change in the common time effect, at
an annualized rate) as the solid line with + marks. Clearly, including αt does make some difference in
the results although the overall patterns do not differ dramatically. For example, the returns to experience
continue to show considerable variation over time. However, one can see that the pattern of the returns to

18

experience is distinctly less procyclical in the bottom panel than in the top. Most notably a dip in the
returns to experience in 1992 just after the 1990-91 recession is clearly visible in the top pattern but
disappears in the bottom.
In the first two rows of Table 1 we formally test whether the time patterns apparent in the figures are
statistically significant by performing Wald tests for whether the parameter values are constant across all
months of the data.19 The columns refer to the parameter being tested and the rows refer to whether the
model contains the common time effect. The table entries show the p-values from the Wald tests.
Whether or not we include the common time effect, we strongly reject the null hypothesis that the return
to experience is constant over time. We also reject the null that the return to job to job switches is
constant over time. However, despite the notable movements shown in Figure 3a, we actually do not
reject that the job to non-employment to job return is constant, although the p-value in the second model
is 0.0518 which is very close to a marginal rejection of constancy.
In Table 2, we test the cyclicality of the parameter values by taking each of the time series of monthly
estimates and regressing them on a constant, a time trend, and the monthly unemployment rate. The
entries in the table show the coefficient on the unemployment rate and standard error from the regression.
In our model without the common time effect, we find that the return to experience is highly pro-cyclical.
A one percent increase in the unemployment rate is associated with a one percent annual decline in the
return to experience. This result is highly significant. The coefficients on job to job and job to nonemployment to job rates are both procyclical: a one percent increase in the unemployment rate is
associated with a half point percent decline in wage growth for both job to job and job to nonemployment to job switchers. However, the pro-cyclical patterns for job switchers are not statistically
significant.
When we include the common time effect, which is strongly cyclical, we find that even the return to
experience is not cyclical. This is not purely due to an increase in standard errors. The point estimate on
cyclicality for the returns to schooling in Table 2 falls from -0.0113 to -0.0003 after we control for ∆lαt .

19

The results strongly suggest that the pro-cyclicality of β t was due to pro-cyclicality associated with
common component to wage levels rather than in the returns to experience. Interpreting this in terms of
the economics framework in the methodological section, this suggests that the human capital rental rate
appears to be procyclical while the human capital production function is not. When interpreted in this
way, this result seems quite reasonable, the demand for human capital rises during a boom due to
increased productivity. This suggests that this component is actually a transitory component of wage
levels rather than permanent wage growth. We view this as an important finding. Because of the
problems with this specification discussed in Section 3, results from this specification should be taken
with some caution. However, the fact that including this variable in the regression makes such a
difference suggests that at the very least it is not simply noise.
In Figure 3b we show the parameter estimates when we estimate the model separately by education
groups. Figure 3c shows parameter estimates for men and women. Both figures refer to results from the
first model without αt (that is to say, equation 2). The top left panel of Figures 3b and 3c show the
estimated return to experience. Although there is a lot of variability in these series, both across time and
across demographic groups, there is no strong secular change in the return to experience. Looking at the
top left panel of Figure 3b, it appears as if the trends in return to experience may differ by education
groups with larger trends for the more educated. We examined this more in depth. While the point
estimates do suggest stronger trends for the more educated, the standard errors are large enough that one
can not say much about this trend with confidence. Furthermore, for all demographic groups (except
college), the return to experience is about 4 percent. In other words, an additional year of experience,
holding all else constant, increases wages by about 4 percent. The return to experience for college
graduates is about 5 percent. The top right and bottom left panels of Figures 3b and 3c also show the
average wage change associated with a job to job change and a job to non-employment to job change,
respectively.

20

The most dramatic difference across groups is that the coefficients for dropouts are particularly procyclical —both the return to experience and the return to job to non employment to job changes appear to
be more pro-cyclical than for other groups.
When we account for the common time effect αt , results are very similar to the results in Figures 3b
and 3c, so we do not present those results. Accounting for αt slightly reduces the estimated return to
experience for high school graduates and drop-outs (although it is still estimated at over 3% for both
groups). Also, the estimated fall in the return to experience for college graduates and high school
graduates largely vanishes. As in figure 3a, the return to experience appears to be less procyclical after
accounting for the common time effect, although due to the noisier results the pattern is less transparent.
To summarize, all three of the returns to labor market transitions vary considerably over time and all
exhibit some pro-cyclicality. However, only the return to experience is related to the business cycle in a
statistically significant fashion. Once we control for the common time effect, however, the return to
experience is no longer pro-cyclical.

Patterns in Labor Market Transitions
In the previous section we examined how the returns to labor market transitions changed over time
and examined the cyclical properties of our parameter estimates. However, wage growth can also vary
over time because of changes in the rates of these transitions irrespective of their returns. For example,
more people are working and accumulating experience during booms. With that goal in mind we now
document trends in labor market transitions over the last 20 years. Recall that we examine three types of
labor market transitions in our model: staying on the same job, job to job transitions and job to nonemployment to job transitions. Also recall that because the SIPP panels are short, we cannot directly
observe all job to non-employment to job transitions, as we do not know if individuals moving from nonemployment to employment were ever previously employed. Therefore, in order to measure these
transition rates with our SIPP data, we estimate the three transition probabilities described in section 3:
employment to non-employment, non-employment to employment, and job to job. We then use these

21

transition probabilities to simulate the labor market transitions for our model. Clearly, the job to job
transition rate is directly related to the probability of a job to job switch. However, the job to nonemployment to job switches depend on two transition probabilities, the transition into non-employment
and then the subsequent transition to employment. Experience accumulation also depends on both the job
to non-employment rate and the non-employment to employment rate.
We calculate the underlying transition rates that determine our labor market transitions annually, for
the combined group of dropouts and high school graduates. These are shown in Figure 4a. The patterns
show considerable variability over time. These movements are strongly statistically significant as can be
seen in the bottom row of Table 1 where we show that we strongly reject that the transition rates are
constant over time. In the bottom row of Table 2 we also find that these rates are all closely related to the
business cycle (in the direction that one would expect). As we mentioned earlier, this is also consistent
with the findings of Eva Nagypal (2004) and Robert Shimer (2005).
We present the transition rates by education group in Figure 4b and by gender in Figure 4c. The
results are fairly similar across groups with a few notable exceptions. Non-employment to employment
transition rates are higher for men and for more educated workers. Employment to non-employment
transition rates are lower for men and for more educated workers. Therefore, the higher employment
rates of men and the educated are the result of both higher non-employment to employment transition
rates and lower employment to non-employment transition rates. Furthermore, non-employment to
employment transitions for less educated workers are more pro-cyclical than for more educated workers.
Another notable difference is a much stronger decline in job to job transitions for high school dropouts
than for the other education groups.
While all three transition rates appear to be trending downward somewhat, all three are highly
cyclical as one would expect. The job to job rate and the non-employment to job rate are highly procyclical while the job to non-employment rate is countercyclical.
We use the estimated transition rates and the simulation model to calculate the probability that
someone is employed, making a job to change, or a job to non-employment to job change for each month

22

during our sample period. For the most part, results from the model are unsurprising, so we do not show
graphs of the results. The employment rate, which we use for measuring the amount of accumulated
experience, is pro-cyclical. For our base sample of high school graduates and drop-outs, the employment
rate rises from 69% in 1984 to 73% in 1989, declines to 69% in 1993, rises to 75% in 1999, then declines
back to 70% in 2003. These participation rates line up closely with values from the CPS.20 The job to
non-employment to job rate is slightly pro-cyclical, falling from 2.1% per month in the mid-1980s to
1.5% in 1992, rising to 1.7% in 1996, falling to 1.5% in 1999, then rising to 1.7% in 2003. Recall that the
job to non-employment to job rate is a function of the job to non-employment rate (which is
countercyclical) and the non-employment to job rate (which is pro-cyclical), so the resulting series is
somewhat acyclical. Lastly, the simulated job to job transition rate unsurprisingly looks like the
estimated job to job transition rate.

Decomposing Changes in Wage Growth Over Time
Thus far we have shown that there have been substantial changes over time in labor market transition
rates and in the wage changes associated with these transitions. In this section we decompose overall
wage growth into its various components using the Oaxaca decomposition approach described in equation
(9). This method basically attempts to ask the following question: what would the time series of wage
growth look like if we simply held the transition rates constant at their average value over the whole time
period so that all wage growth was only due to the changing returns? The result of this experiment would
give us a sense of how much of the observed pattern of wage growth was due to changes in transition
rates versus changes in returns.
In Figure 5 we graph the results of the decomposition described in equation (9) converting our
monthly estimates into annual units. The top panel presents the decomposition of the change in wage
growth over time ( Ε[∆1ωit ] ) while the bottom panel presents the decomposition of wage changes net of
the common time component ( Ε[∆1ωit − ∆1αt ] ). In both panels the solid line represents the overall
predicted wage growth ( G ′tΧt ). The dashed line presents the component that only allows the coefficients

23

(returns) to change G ′tΧ while the dotted line presents the remainder term ( G ′t ( Χt − Χ ) ). Clearly the
dashed line explains virtually all of the change in the lifecycle wage growth. This result is also robust
across subgroups (not shown). In short, changes in the coefficients explain almost all of the variability in
wages, and changes in the transition rates explain essentially none of it.
Another perhaps surprising aspect of this is that we do not find that trends in the amount of labor
force experience (due to labor supply) are important. We looked at this more closely for both men and
women. For this sample there is essentially no trend in labor force experience for either of these groups.
We find this result extremely surprising. As expected experience and job to job transitions are
associated with wage gains while non-employment spells are associated with wage losses. Also, as
expected, job to job and job finding rates are highly pro-cyclical while job to non-employment rates are
counter cyclical. However, the magnitude of these effects in explaining the variation in overall wage
growth is miniscule as is clearly evident in Figure 5.
We next decompose the dashed line in Figure 5 ( G ′tΧ ) into its various parts: a component related to
the return to experience, a component related to job to job switches, and a component related to job to
unemployment to job switches. We show this for both of our wage growth models in the two panels in
Figure 6. The solid line in each panel reproduces the value of G ′tΧ from Figure 5. We then present the
three different components that contribute to it. Very clearly, the coefficient on experience is chiefly
responsible for the main result. In short, the great majority of the variability in wage growth over time
comes from one source: variability in the returns to experience.
Another interesting finding is also apparent in Figure 6. While the level of wage growth is entirely
accounted for by the return to experience, job to job transitions lead to positive effects while job to nonemployment to job transitions lead to negative effects and these two roughly offset each other. In most
years wage growth associated with job changes accounts for less than 10% of wage growth. Thus, we
find that job changes are less important for understanding early career wage growth than does Robert
Topel and Michael Ward (1992), who find 30% of early career wage growth occurs at job to job

24

transitions. Gadi Barlevy (2005) finds that job to job transitions are more important for wage growth than
what we find, but are less important for wage growth than what Robert Topel and Michael Ward find.
Further research is necessary to better understand whether differences in results are attributable to
differences in data (ours is more recent and for less educated workers), or differences in methodology.
Ideally, we would like to understand why the return to experience has changed over time but
unfortunately we see no obvious explanation. We leave further exploration of this result to future
research.

6.

Conclusions
This chapter analyzes the changing patterns of early career wage growth for less skilled workers over

the last twenty years. We find that wage progression has varied considerably over this period. Wage
growth for young individuals averaged about 4.5 percent over this period ranging from as high as 6
percent to as low as 3 percent.
We develop a model of wage changes in order to identify the key determinants of lifecycle wage
growth for younger workers and how these factors have influenced the changing pattern of wage growth.
In our first specification, in which we do not attempt to account for the common time effect on wages
shared by all workers, we find that the returns to experience are highly variable and pro-cyclical. In our
second specification which incorporates a common time effect, we find no relationship between the return
to experience and the business cycle. Instead, we find that most of the pro-cyclicality of wage growth
comes from the common time effect. Nevertheless, even in this second model the returns to experience
still exhibits a fair degree of variation over time.
Because our strategy for identifying the common time effect is problematic, we cannot be sure if it is
the common time effect or the return to experience that causes the procyclicality of wages during our
sample period. To the extent that it is the common time effect that is responsible for the pro-cyclicality in
wages, the results are straightforward to interpret. Improvements in technology and increases in the
amount of capital per worker imply that the productivity of all workers increases, regardless of experience

25

level. This makes it a good time to hire, even if wages of workers are being bid up. Thus, wages of
continuing workers rises.
We then examine wage gains associated with different labor market transitions in order to identify the
relative importance of wage growth on the job (which we interpret as the return to experience, plus,
potentially a common time component) and wage growth when moving across jobs (which we interpret as
the change in job match quality). Surprisingly we find that virtually all of the change in wage growth
over time is accounted for by wage growth on the job. Although we do find that experience accumulation
and job changes are related to the business cycle, the magnitude of the contributions to overall wage
growth from these sources is negligible.
Although there is variability in the return to experience, it has averaged a healthy 4 percent over our
sample period and generally speaking, does not vary much by gender or education level. The fact that the
returns to experience have moved in ways unrelated to the business cycle is an interesting finding that
deserves more attention. We do not have any obvious explanation of this result and thus are hesitant to
make policy prescriptions based on this finding. Future research both to confirm our results and to delve
deeper into this issue is clearly needed.
Taken as a whole, our results suggest that the business cycle plays a surprisingly small role in
lifecycle wage growth and that policymakers should not be overly concerned that workers who enter the
labor market during recessions will suffer slower wage growth over their careers. While it is true that
workers who are in the labor market during a downturn will experience longer unemployment spells and
fewer job to job transitions, this will have relatively little effect on wage growth over the first ten years of
the career. Overall, our results suggests that experience accumulation is potentially an important pathway
out of poverty and so attempts by policymakers to encourage work do not seem to be misplaced. In
contrast, we do not find job switching to be an important explanation for early career wage growth, at
least on average. Nonetheless, policies that produce particularly high quality matches may still be
desirable.

26

References
Abraham, Kathrine, and John Haltiwanger. 1995. “Real Wages and the Business Cycle.” Journal of
Economic Literature 33: 215-264.
Andersson, Fredrik, Harry Holzer, and Julia Lane. 2005. Moving Up or Moving On. New York: Russell
Sage Foundation.
Antel, John. 1991. “The Wage Effects of Voluntary Labor Mobility with and without Intervening Nonemployment.” Industrial and Labor Relations Review 44: 299-306.
Barlevy, Gadi. 2005. “Estimating Models of On the Job Search Using Record Statistics.” Manuscript.
Ben Porath, Yoram. 1967. “The Production of Human Capital and the Life Cycle of Earnings.” Journal of
Political Economy 75(4), pt. 1: 352-65.
Beaudry, Paul, and John Dinardo. 1991. “The Effect of Implicit Contracts on the Movement of Wages
over the Business Cycle: Evidence from Micro Data.” Journal of Political Economy 99 no. 4: 665-688.
Burtless, Gary 1994. “Employment Prospects for Welfare Recipients” In The Work Alternative, Welfare
Reform and the Realities of the Job Market, edited by Demetra Smith Nightingale and Robert H.
Haveman. Washington, D.C.: Urban Institute Press.
Connolly, Helen and Peter Gottschalk. “Returns to Tenure and Experience Revisited-Do Less Educated
Workers Gain Less from Work Experience?” Unpublished paper. Boston College, Chestnut Hill.
Davis,Steven, John Haltiwanger, and Scott Schuh. 1996. Job Creation and Destruction. Cambridge : MIT
Press.
Devine, Theresa, and Nicholas Kiefer. 1991. Empirical Labor Economics. Oxford: Oxford University
Press.
Farber, Henry. 1994. “The Analysis of Interfirm Worker Mobility.” Journal of Labor Economics 12: 554593.
Farber, Henry. 2005. “What Do We Know about Job Loss in the United States? Evidence from the
Displaced Worker’s Survey, 1984-2004.” Economic Perspectives, 2nd Quarter 2005: 13-28.
Flinn, Christopher. 1986. “Wages and Job Mobility of Young Workers.” Journal of Political Economy
94: S88-S110.
Flinn, Christopher, and James Heckman. 1982. “New Methods for Analyzing Structural Models of Labor
Force Dynamics.” Journal of Econometrics: 115-168.
Gladden, Tricia, and Christopher Taber 2000. “Wage Progression Among Low Skilled Workers.” In
Finding Jobs: Work and Welfare Reform, edited by Card and Blank. New York: Russell Sage Foundation.
Gladden, Tricia, and Christopher Taber. 2004. “The Relationship Between Wage Levels and Wage
Growth.” Unpublished paper. Northwestern University, Evanston.
Gladden, Tricia, and Christopher Taber. 2006. “Turnover and Wage Growth in the Transition from
School to Work,” Unpublished paper, Northwestern University, Evanston.
Grogger, Jeffrey. “Welfare Reform, Returns to Experience, and Wages.” Unpublished paper. University
of Chicago, Chicago.

27

Heckman, J., Lochner, L., and C. Taber. 1998. “Explaining Rising Wage Inequality: Explorations with a
Dynamic General Equilibrium Model of Labor Earnings with Heterogeneous Agents,” Review of
Economic Dynamics.
Holzer, Harry, and Robert Lalonde. 2000. “Job Change and Job Stability Among Less Skilled Young
Workers.” In Finding Jobs: Work and Welfare Reform, edited by Card and Blank. New York: Russell
Sage Foundation.
Klerman, Jacob, and Lynn Karoly. 1994. “Young Men and the Transition to Stable Employment.”
Monthly Labor Review: 31-48.
Corcoran, Mary, and Susanna Loeb. 1999. “Welfare, Work Experience, and Economic Self-Sufficiency.”
Unpublished paper. University of Michigan
Loprest, Pamela. 1992. “Gender Differences in Wage Growth and Job Mobility,” American Economic
Review: 526-532.
Mincer, J. 1962. “On the Job Training: Costs, Returns and Some Implications.” Journal of Political
Economy 70, Supplement, #5, part 2: 50-79.
Mincer, Jacob, and Boyan Jovanovic. 1981. “Labor Mobility and Wages.” In Studies in Labor Markets,
edited by Sherwin Rosen. Chicago: Chicago University Press.
Moffitt, Robert, and Anuradha Rangarajan. 1989. “The Effect of Transfer Programmes on Work Effort
and Human Capital Formation: Evidence from the US,” in The Economics of Social Security, edited by
Andrew Dilnot and Ian Walker. Oxford: Oxford University
Press.
Nagypal, Eva. 2004. “Worker Reallocation over the Business Cycle: The Importance of Job-to-Job
Transitions.” Working paper. Northwestern University, Evanston.
Neumark, David. 2002. “Youth Labor Markets in the US: Shopping Around Vs. Staying Put.” Review of
Economics and Statistics 84(3): 462-482.
Oreopolous, Phil, Till von Wachter and Andrew Heisz. 2006. “The Short- and Long-Term Career Effects
of Graduating in a Recession: Hysteresis and Heterogeneity in the Market for College Graduates.”
Unpublished Manuscript, Columbia University.
Shimer, Robert. 2005. “The Cyclicality of Hires, Separations, and Job-to-Job Transitions,” Unpublished
paper. University of Chicago, Chicago.
Topel, Robert and Michael Ward. 1992. “Job Mobility and the Careers of Young Men.”
Quarterly Journal of Economics 108: 439-479.
Wolpin, Kenneth. 1992. “The Determinants of Black-White Differences in Early Employment Careers:
Search, Layoffs, Quits, and Endogenous Wage Growth.” Journal of
Political Economy 100: 535-560.

28

Table 1: Tests of the constancy of parameter estimates and transition rates
p-values on Parameters from Wage Growth Model
Coefficient on
Experience

Coefficient on Job to
Job Transitions

Model without α

0.003

0.035

Coefficient on Job to
Non-employment to
Job Transitions
0.227

Model with α

0.000

0.005

0.052

p-values on Labor Market Transition Rates
Job to Job
0.0000

Job to Nonemployment
0.0000

Non-employment to
Job
0.0000

Note: Table entries show the probability values for the null hypothesis of parameter constancy over time
against the alternative of time varying parameters

Table 2: Tests of the cyclicality of parameter estimates and transition rates
Regression of Parameters from Wage Growth Model on Unemployment Rate
Coefficient on
Experience

Change in Wage at
Job to Job Transitions

Change in Wage at
Job to Nonemployment to Job
Transitions
Model without α
-0.0113
-0.0051
-0.0055
(0.0033)*
(0.0055)
(0.0069)
Model with α
-0.0003
-0.0053
-0.0018
(0.0079)
(0.0055)
(0.0061)
Regression of Job Transition Probabilities on Unemployment Rate
Job to Job

Job to Nonemployment
0.0015
(0.0006)*

-0.0024
(0.0006)*

Non-employment to
Job
-0.0064
(0.0015)*

Notes: Table entries show the results of regressions on the monthly unemployment rate. Newey-West
standard errors in parentheses
* Statistically significant at the 5% level

29

Endnotes
1

We thank Dan Aaronson for his help with the CPS data and Phil Doctor for excellent research assistance. We
thank the editors, Peter Gottschalk, and two anonymous referees for comments.
2
For example, wages of high school drop-outs aged 18-28 fell from about $12 an hour 1979 (in 2004 dollars) to
under $10 an hour in 2004.
3
We constructed this profile using the 1996 SIPP panel. We regressed log wages on age dummies and calendar year
dummies for high school graduate and high school dropout men. We then present the predicted value from this
regression in 1996 looking across ages. The profile, then, should be interpreted as the wages of individuals of
different ages in a single year.
4
It is important to distinguish our discussion of changes in wages for individuals from the extensive literature on the
cyclicality of aggregate wage levels.
5
Specifically, we estimate the percent increase in the hourly wage for individuals who were working in the survey
wave, and were also working in the previous Outgoing Rotation Group 12 months prior.
6
Note that sample members in Figure 2 age one year between the two time periods we observe them. In other
words, we are measuring wage growth between ages 18 and 19, 19 and 20, and so on for the same people between
adjacent years. Furthermore, Figure 1 shows that wages increase rapidly with age. So although there has been little
aggregate change in wage levels for low skilled workers over this time period, the wages of continuously employed
young workers grow as they age.
7
In the augmented model that accounts for the common component to wages, we find that this component explains
a large portion of wage growth. This is not surprising given previous work on the changing wage structure and the
variation in wages across the business cycle. Nonetheless, wage growth net of this component, is virtually entirely
explained by changes in the return to experience.
8
James Heckman, Lance Lochner, and Christopher Taber (1998) provide a more recent version of this basic model.
9
Robert Shimer (2005) claims that the differences come from two sources. First, Steven Davis, John Haltiwanger,
and Scott Schuh (1996) use manufacturing sector data, which may not be representative of the economy as a whole.
Second, they use plant level data instead of individual level data. A plant can destroy jobs by merely not hiring new
workers. If firms quit hiring new workers during a recession, jobs will be destroyed as workers quit (although the
job separation rate may not necessarily rise), but the hiring rate will fall. Therefore, a countercyclical job destruction
rate is consistent with an acyclical job separation rate.
10
Equation (2) can be derived from the following model of wage levels:
t

ω it = θ i + ∑ β s ( Α is − Α is − 1) + η ij + ε it
it

s=0

where θ i is an individual specific fixed effect that is potentially correlated with experience Αis and match quality
ηijit . First differencing this equation and assuming that βs changes sufficiently slowly that β t ≈ β t − l yields
equation (2).
11
In a standard human capital model one would expect the returns to experience to be countercyclical as the cost of
investing in human capital is lower during a recession.
12
See Tricia Gladden and Christopher Taber (2006) for an attempt to measure the distribution of ηijit .
13
This is essentially an example of the fundamental problem of separating time, age, and cohort effects. For stayers
age, time, and experience are perfectly collinear.
14
For details on the simulation please see the working paper version of this study available at
www.faculty.econ.northwestern.edu/faculty/taber.
15
In the CPS, job to job transitions can only be identified beginning in 1994 (Bruce Fallick and Charles Fleischman,
2001).
16
For the 1996 and 2001 SIPPs job changes are only identified across waves.
17
The fact that some months contain 5 weeks rather than 4 sometimes leads to spurious wage changes within a
wave. For these cases we use 4.3 weeks rather than the actual weeks to calculate the wage. In the 1996 and 2001
SIPP panels we do not know weeks worked in a month by employer. Adjusting the pre-1996 SIPP data to
correspond to this measure of weeks has almost no effect on our results.
18
We should point out that Farber estimates the earnings loss associated with being a displaced worker, whereas our
measure is the wage loss for job to non-employment to job transitions. The associated job loss may be voluntary, or
the job loss may be for cause. Furthermore, we measure wages whereas Farber measures earnings. Lastly, we
measure wage losses for a younger group of workers, who have not had the time to find good matches and have not

30

had time to gain firm-specific human capital. Nevertheless, we both find that earnings losses associated with job loss
diminished in size during the 1990s.
19
Since we are using a kernel smoother to present the results in the figures, this can make parameter estimates
appear more stable over time than the actually are. Because we are using a two stage procedure to estimate the mean
wage changes for movers, calculation of standard errors is not straight forward since one must correct for the first
stage estimation of βt. We allow for this estimation error by treating the full model as one large GMM system and
then perform Wald tests for constancy of parameter values across all months.
20
Employment rates in the CPS tend to be about 2% lower than in our simulated data in every year. However, this
difference does not change over time, so both series show almost identical cyclical fluctuations.

31

14

13

12

Predicted Wages

11

10

9

8

7

6
15

20

25

30

35

40

45

50

Age

Figure 1: Log Wage Profile, Male, No College

1

55

60

0.12

0.1

annual wage growth

0.08

0.06

0.04

0.02

0

−0.02

−0.04
1980

1985

1990

1995

2000

year

Solid Line All Workers
Dashed Line High School Dropouts
Vertical Line High School Graduates
Solid Line with O Beyond High School

Figure 2: Wage Growth 1980-2004, CPS by Education Group

2

2005

Without Common Time Effect (αt )
0.06

0.04

Coefficient

0.02

0

−0.02

−0.04

−0.06
1984

1986

1988

1990

1992

1994
year

1996

1998

2000

2002

2004

1998

2000

2002

2004

With Common Time Effect (αt )
0.06

0.04

Coefficient

0.02

0

−0.02

−0.04

−0.06
1984

1986

1988

1990

1992

1994
year

1996

Solid Line Experience Change
Dashed Line Job to Job Transitions
Vertical Line Job to Nonemployment to Job Transitions
Solid Line with + Intercept change
3
Figure 3a: Coefficients Over Time (Gt ), ages 18-28, no college

Experience Change

Job to Job Changes

0.08

0.06

0.07
0.05
0.06
0.04

Coefficient

Coefficient

0.05

0.04

0.03

0.03

0.02

0.02
0.01
0.01
0
0

−0.01
1984

1986

1988

1990

1992

1994
year

1996

1998

2000

2002

2004

−0.01
1984

1986

1988

1990

1992

1994
year

Job to Nonemployment to Job Changes
0.04

0.02

Coefficient

0

−0.02

−0.04

−0.06

−0.08

−0.1
1984

1986

1988

1990

1992

1994
year

1996

1998

2000

2002

2004

Dashed Line High School Drop Outs
Solid Line High School Graduates
Vertical Line College Attenders
4
Figure 3b: Coefficients Over Time (Gt ), by Education Group

1996

1998

2000

2002

2004

Experience Change

Job to Job Change

0.08

0.06

0.07

0.05

0.06
0.04

Coefficient

Coefficient

0.05

0.04

0.03

0.02

0.03
0.01
0.02

0

0.01

0
1984

1986

1988

1990

1992

1994
year

1996

1998

2000

2002

2004

−0.01
1984

1986

1988

1990

Job to Nonemployment to Job Changes
0.03

0.02

0.01

Coefficient

0

−0.01

−0.02

−0.03

−0.04

−0.05

−0.06
1984

1986

1988

1990

1992

1994
year

1996

1998

2000

2002

2004

Solid Line Men
Dashed Line Women

Figure 3c: Coefficients Over Time (Gt ), by Sex
5

1992

1994
year

1996

1998

2000

2002

2004

0.09

0.08

Transition Probabilities

0.07

0.06

0.05

0.04

0.03

0.02
1984

1986

1988

1990

1992

1994
year

1996

1998

2000

2002

Solid Line Nonemployment to Employment Transition
Dashed Line Job to Nonemployment Transition
Vertical Line Job to Job Transitions

Figure 4a: Transition Probabilities, 18-28 year olds, no college

6

2004

Nonemployment to Job

Job to Nonemployment

0.05

0.11

0.045

0.1

0.04
Transition Probabilities

0.055

0.12

Transition Probabilities

0.13

0.09

0.08

0.035

0.03

0.07

0.025

0.06

0.02

0.05

0.015

0.04
1984

1986

1988

1990

1992

1994
year

1996

1998

2000

2002

2004

1998

2000

2002

0.01
1984

1986

1988

1990

1992

2004

Job to Job
0.045

Transition Probabilities

0.04

0.035

0.03

0.025

0.02
1984

1986

1988

1990

1992

1994
year

1996

Solid Line High School Graduates
Dashed Line High School Drop Outs
Vertical Line College Attenders

7
Figure 4b: Transition Probabilities, by Education Group

1994
year

1996

1998

2000

2002

2004

Nonemployment to Job

Job to Nonemployment

0.16

0.04

0.14
0.035

Transition Probabilities

Transition Probabilities

0.12

0.1

0.03

0.025

0.08

0.02
0.06

0.04
1984

1986

1988

1990

1992

1994
year

1996

1998

2000

2002

2004

1998

2000

2002

0.015
1984

1986

1988

2004

Job to Job
0.045

Transition Probabilities

0.04

0.035

0.03

0.025

0.02
1984

1986

1988

1990

1992

1994
year

1996

Solid Line Men
Dashed Line Women

Figure 4c: Transition Probabilities, by Sex
8

1990

1992

1994
year

1996

1998

2000

2002

2004

Without Common Time Effect (αt )
0.06

0.05

Prediced Wage Growth

0.04

0.03

0.02

0.01

0

−0.01
1984

1986

1988

1990

1992

1994
year

1996

1998

2000

2002

2004

1998

2000

2002

2004

With Common Time Effect (αt )
0.04

0.035

0.03

Prediced Wage Growth

0.025

0.02

0.015

0.01

0.005

0

−0.005
1984

1986

1988

1990

1992

1994
year

1996

Solid Line Overall Predicted Wage Growth
Dashed Line Allowing Only Slope Coefficients (Gt ) to Change
Vertical Line Residual Due to Factor (Xt ) Changes

Figure 5: Predicted Wage, Decomposed into Subcomponents
9

Without Common Time Effect (αt )
0.06

0.05

Prediced Wage Growth

0.04

0.03

0.02

0.01

0

−0.01

−0.02
1984

1986

1988

1990

1992

1994
year

1996

1998

2000

2002

2004

1998

2000

2002

2004

With Common Time Effect (αt )
0.04

0.03

Prediced Wage Growth

0.02

0.01

0

−0.01

−0.02
1984

1986

1988

1990

1992

1994
year

1996

Solid Line Overall Growth Due to Coefficients
Dashed Line Wage Growth on Job
Vertical Line Job to Job Transitions
Solid Line with + Job to Nonemployment to Job Transitions
10
Figure 6: Decomposition of Wage Growth: Coefficients

Without Common Time Effect (αt )
0.06

0.05

Prediced Wage Growth

0.04

0.03

0.02

0.01

0

−0.01
1984

1986

1988

1990

1992

1994
year

1996

1998

2000

2002

2004

1998

2000

2002

2004

With Common Time Effect (αt )
0.04

0.035

0.03

Prediced Wage Growth

0.025

0.02

0.015

0.01

0.005

0

−0.005
1984

1986

1988

1990

1992

1994
year

1996

Solid Line Overall Predicted Wage Growth
Dashed Line Allowing Only Slope Coefficients (Gt ) to Change
Vertical Line Residual Due to Factor (Xt ) Changes

Figure 6: Predicted Wage, Decomposed into Subcomponents
10

Without Common Time Effect (αt )
0.06

0.05

Prediced Wage Growth

0.04

0.03

0.02

0.01

0

−0.01

−0.02
1984

1986

1988

1990

1992

1994
year

1996

1998

2000

2002

2004

1998

2000

2002

2004

With Common Time Effect (αt )
0.04

0.03

Prediced Wage Growth

0.02

0.01

0

−0.01

−0.02
1984

1986

1988

1990

1992

1994
year

1996

Solid Line Overall Growth Due to Coefficients
Dashed Line Wage Growth on Job
Vertical Line Job to Job Transitions
Solid Line with + Job to Nonemployment to Job Transitions
11
Figure 7: Decomposition of Wage Growth: Coefficients

Working Paper Series
A series of research studies on regional economic issues relating to the Seventh Federal
Reserve District, and on financial and economic topics.
Outsourcing Business Services and the Role of Central Administrative Offices
Yukako Ono

WP-02-01

Strategic Responses to Regulatory Threat in the Credit Card Market*
Victor Stango

WP-02-02

The Optimal Mix of Taxes on Money, Consumption and Income
Fiorella De Fiore and Pedro Teles

WP-02-03

Expectation Traps and Monetary Policy
Stefania Albanesi, V. V. Chari and Lawrence J. Christiano

WP-02-04

Monetary Policy in a Financial Crisis
Lawrence J. Christiano, Christopher Gust and Jorge Roldos

WP-02-05

Regulatory Incentives and Consolidation: The Case of Commercial Bank Mergers
and the Community Reinvestment Act
Raphael Bostic, Hamid Mehran, Anna Paulson and Marc Saidenberg

WP-02-06

Technological Progress and the Geographic Expansion of the Banking Industry
Allen N. Berger and Robert DeYoung

WP-02-07

Choosing the Right Parents: Changes in the Intergenerational Transmission
of Inequality  Between 1980 and the Early 1990s
David I. Levine and Bhashkar Mazumder

WP-02-08

The Immediacy Implications of Exchange Organization
James T. Moser

WP-02-09

Maternal Employment and Overweight Children
Patricia M. Anderson, Kristin F. Butcher and Phillip B. Levine

WP-02-10

The Costs and Benefits of Moral Suasion: Evidence from the Rescue of
Long-Term Capital Management
Craig Furfine

WP-02-11

On the Cyclical Behavior of Employment, Unemployment and Labor Force Participation
Marcelo Veracierto

WP-02-12

Do Safeguard Tariffs and Antidumping Duties Open or Close Technology Gaps?
Meredith A. Crowley

WP-02-13

Technology Shocks Matter
Jonas D. M. Fisher

WP-02-14

Money as a Mechanism in a Bewley Economy
Edward J. Green and Ruilin Zhou

WP-02-15

1

Working Paper Series (continued)
Optimal Fiscal and Monetary Policy: Equivalence Results
Isabel Correia, Juan Pablo Nicolini and Pedro Teles

WP-02-16

Real Exchange Rate Fluctuations and the Dynamics of Retail Trade Industries
on the U.S.-Canada Border
Jeffrey R. Campbell and Beverly Lapham

WP-02-17

Bank Procyclicality, Credit Crunches, and Asymmetric Monetary Policy Effects:
A Unifying Model
Robert R. Bliss and George G. Kaufman

WP-02-18

Location of Headquarter Growth During the 90s
Thomas H. Klier

WP-02-19

The Value of Banking Relationships During a Financial Crisis:
Evidence from Failures of Japanese Banks
Elijah Brewer III, Hesna Genay, William Curt Hunter and George G. Kaufman

WP-02-20

On the Distribution and Dynamics of Health Costs
Eric French and John Bailey Jones

WP-02-21

The Effects of Progressive Taxation on Labor Supply when Hours and Wages are
Jointly Determined
Daniel Aaronson and Eric French

WP-02-22

Inter-industry Contagion and the Competitive Effects of Financial Distress Announcements:
Evidence from Commercial Banks and Life Insurance Companies
Elijah Brewer III and William E. Jackson III

WP-02-23

State-Contingent Bank Regulation With Unobserved Action and
Unobserved Characteristics
David A. Marshall and Edward Simpson Prescott

WP-02-24

Local Market Consolidation and Bank Productive Efficiency
Douglas D. Evanoff and Evren Örs

WP-02-25

Life-Cycle Dynamics in Industrial Sectors. The Role of Banking Market Structure
Nicola Cetorelli

WP-02-26

Private School Location and Neighborhood Characteristics
Lisa Barrow

WP-02-27

Teachers and Student Achievement in the Chicago Public High Schools
Daniel Aaronson, Lisa Barrow and William Sander

WP-02-28

The Crime of 1873: Back to the Scene
François R. Velde

WP-02-29

Trade Structure, Industrial Structure, and International Business Cycles
Marianne Baxter and Michael A. Kouparitsas

WP-02-30

Estimating the Returns to Community College Schooling for Displaced Workers
Louis Jacobson, Robert LaLonde and Daniel G. Sullivan

WP-02-31

2

Working Paper Series (continued)
A Proposal for Efficiently Resolving Out-of-the-Money Swap Positions
at Large Insolvent Banks
George G. Kaufman

WP-03-01

Depositor Liquidity and Loss-Sharing in Bank Failure Resolutions
George G. Kaufman

WP-03-02

Subordinated Debt and Prompt Corrective Regulatory Action
Douglas D. Evanoff and Larry D. Wall

WP-03-03

When is Inter-Transaction Time Informative?
Craig Furfine

WP-03-04

Tenure Choice with Location Selection: The Case of Hispanic Neighborhoods
in Chicago
Maude Toussaint-Comeau and Sherrie L.W. Rhine

WP-03-05

Distinguishing Limited Commitment from Moral Hazard in Models of
Growth with Inequality*
Anna L. Paulson and Robert Townsend

WP-03-06

Resolving Large Complex Financial Organizations
Robert R. Bliss

WP-03-07

The Case of the Missing Productivity Growth:
Or, Does information technology explain why productivity accelerated in the United States
but not the United Kingdom?
Susanto Basu, John G. Fernald, Nicholas Oulton and Sylaja Srinivasan

WP-03-08

Inside-Outside Money Competition
Ramon Marimon, Juan Pablo Nicolini and Pedro Teles

WP-03-09

The Importance of Check-Cashing Businesses to the Unbanked: Racial/Ethnic Differences
William H. Greene, Sherrie L.W. Rhine and Maude Toussaint-Comeau

WP-03-10

A Firm’s First Year
Jaap H. Abbring and Jeffrey R. Campbell

WP-03-11

Market Size Matters
Jeffrey R. Campbell and Hugo A. Hopenhayn

WP-03-12

The Cost of Business Cycles under Endogenous Growth
Gadi Barlevy

WP-03-13

The Past, Present, and Probable Future for Community Banks
Robert DeYoung, William C. Hunter and Gregory F. Udell

WP-03-14

Measuring Productivity Growth in Asia: Do Market Imperfections Matter?
John Fernald and Brent Neiman

WP-03-15

Revised Estimates of Intergenerational Income Mobility in the United States
Bhashkar Mazumder

WP-03-16

3

Working Paper Series (continued)
Product Market Evidence on the Employment Effects of the Minimum Wage
Daniel Aaronson and Eric French

WP-03-17

Estimating Models of On-the-Job Search using Record Statistics
Gadi Barlevy

WP-03-18

Banking Market Conditions and Deposit Interest Rates
Richard J. Rosen

WP-03-19

Creating a National State Rainy Day Fund: A Modest Proposal to Improve Future
State Fiscal Performance
Richard Mattoon

WP-03-20

Managerial Incentive and Financial Contagion
Sujit Chakravorti and Subir Lall

WP-03-21

Women and the Phillips Curve: Do Women’s and Men’s Labor Market Outcomes
Differentially Affect Real Wage Growth and Inflation?
Katharine Anderson, Lisa Barrow and Kristin F. Butcher

WP-03-22

Evaluating the Calvo Model of Sticky Prices
Martin Eichenbaum and Jonas D.M. Fisher

WP-03-23

The Growing Importance of Family and Community: An Analysis of Changes in the
Sibling Correlation in Earnings
Bhashkar Mazumder and David I. Levine

WP-03-24

Should We Teach Old Dogs New Tricks? The Impact of Community College Retraining
on Older Displaced Workers
Louis Jacobson, Robert J. LaLonde and Daniel Sullivan

WP-03-25

Trade Deflection and Trade Depression
Chad P. Brown and Meredith A. Crowley

WP-03-26

China and Emerging Asia: Comrades or Competitors?
Alan G. Ahearne, John G. Fernald, Prakash Loungani and John W. Schindler

WP-03-27

International Business Cycles Under Fixed and Flexible Exchange Rate Regimes
Michael A. Kouparitsas

WP-03-28

Firing Costs and Business Cycle Fluctuations
Marcelo Veracierto

WP-03-29

Spatial Organization of Firms
Yukako Ono

WP-03-30

Government Equity and Money: John Law’s System in 1720 France
François R. Velde

WP-03-31

Deregulation and the Relationship Between Bank CEO
Compensation and Risk-Taking
Elijah Brewer III, William Curt Hunter and William E. Jackson III

WP-03-32

4

Working Paper Series (continued)
Compatibility and Pricing with Indirect Network Effects: Evidence from ATMs
Christopher R. Knittel and Victor Stango

WP-03-33

Self-Employment as an Alternative to Unemployment
Ellen R. Rissman

WP-03-34

Where the Headquarters are – Evidence from Large Public Companies 1990-2000
Tyler Diacon and Thomas H. Klier

WP-03-35

Standing Facilities and Interbank Borrowing: Evidence from the Federal Reserve’s
New Discount Window
Craig Furfine

WP-04-01

Netting, Financial Contracts, and Banks: The Economic Implications
William J. Bergman, Robert R. Bliss, Christian A. Johnson and George G. Kaufman

WP-04-02

Real Effects of Bank Competition
Nicola Cetorelli

WP-04-03

Finance as a Barrier To Entry: Bank Competition and Industry Structure in
Local U.S. Markets?
Nicola Cetorelli and Philip E. Strahan

WP-04-04

The Dynamics of Work and Debt
Jeffrey R. Campbell and Zvi Hercowitz

WP-04-05

Fiscal Policy in the Aftermath of 9/11
Jonas Fisher and Martin Eichenbaum

WP-04-06

Merger Momentum and Investor Sentiment: The Stock Market Reaction
To Merger Announcements
Richard J. Rosen

WP-04-07

Earnings Inequality and the Business Cycle
Gadi Barlevy and Daniel Tsiddon

WP-04-08

Platform Competition in Two-Sided Markets: The Case of Payment Networks
Sujit Chakravorti and Roberto Roson

WP-04-09

Nominal Debt as a Burden on Monetary Policy
Javier Díaz-Giménez, Giorgia Giovannetti, Ramon Marimon, and Pedro Teles

WP-04-10

On the Timing of Innovation in Stochastic Schumpeterian Growth Models
Gadi Barlevy

WP-04-11

Policy Externalities: How US Antidumping Affects Japanese Exports to the EU
Chad P. Bown and Meredith A. Crowley

WP-04-12

Sibling Similarities, Differences and Economic Inequality
Bhashkar Mazumder

WP-04-13

Determinants of Business Cycle Comovement: A Robust Analysis
Marianne Baxter and Michael A. Kouparitsas

WP-04-14

5

Working Paper Series (continued)
The Occupational Assimilation of Hispanics in the U.S.: Evidence from Panel Data
Maude Toussaint-Comeau

WP-04-15

Reading, Writing, and Raisinets1: Are School Finances Contributing to Children’s Obesity?
Patricia M. Anderson and Kristin F. Butcher

WP-04-16

Learning by Observing: Information Spillovers in the Execution and Valuation
of Commercial Bank M&As
Gayle DeLong and Robert DeYoung

WP-04-17

Prospects for Immigrant-Native Wealth Assimilation:
Evidence from Financial Market Participation
Una Okonkwo Osili and Anna Paulson

WP-04-18

Individuals and Institutions: Evidence from International Migrants in the U.S.
Una Okonkwo Osili and Anna Paulson

WP-04-19

Are Technology Improvements Contractionary?
Susanto Basu, John Fernald and Miles Kimball

WP-04-20

The Minimum Wage, Restaurant Prices and Labor Market Structure
Daniel Aaronson, Eric French and James MacDonald

WP-04-21

Betcha can’t acquire just one: merger programs and compensation
Richard J. Rosen

WP-04-22

Not Working: Demographic Changes, Policy Changes,
and the Distribution of Weeks (Not) Worked
Lisa Barrow and Kristin F. Butcher

WP-04-23

The Role of Collateralized Household Debt in Macroeconomic Stabilization
Jeffrey R. Campbell and Zvi Hercowitz

WP-04-24

Advertising and Pricing at Multiple-Output Firms: Evidence from U.S. Thrift Institutions
Robert DeYoung and Evren Örs

WP-04-25

Monetary Policy with State Contingent Interest Rates
Bernardino Adão, Isabel Correia and Pedro Teles

WP-04-26

Comparing location decisions of domestic and foreign auto supplier plants
Thomas Klier, Paul Ma and Daniel P. McMillen

WP-04-27

China’s export growth and US trade policy
Chad P. Bown and Meredith A. Crowley

WP-04-28

Where do manufacturing firms locate their Headquarters?
J. Vernon Henderson and Yukako Ono

WP-04-29

Monetary Policy with Single Instrument Feedback Rules
Bernardino Adão, Isabel Correia and Pedro Teles

WP-04-30

6

Working Paper Series (continued)
Firm-Specific Capital, Nominal Rigidities and the Business Cycle
David Altig, Lawrence J. Christiano, Martin Eichenbaum and Jesper Linde

WP-05-01

Do Returns to Schooling Differ by Race and Ethnicity?
Lisa Barrow and Cecilia Elena Rouse

WP-05-02

Derivatives and Systemic Risk: Netting, Collateral, and Closeout
Robert R. Bliss and George G. Kaufman

WP-05-03

Risk Overhang and Loan Portfolio Decisions
Robert DeYoung, Anne Gron and Andrew Winton

WP-05-04

Characterizations in a random record model with a non-identically distributed initial record
Gadi Barlevy and H. N. Nagaraja

WP-05-05

Price discovery in a market under stress: the U.S. Treasury market in fall 1998
Craig H. Furfine and Eli M. Remolona

WP-05-06

Politics and Efficiency of Separating Capital and Ordinary Government Budgets
Marco Bassetto with Thomas J. Sargent

WP-05-07

Rigid Prices: Evidence from U.S. Scanner Data
Jeffrey R. Campbell and Benjamin Eden

WP-05-08

Entrepreneurship, Frictions, and Wealth
Marco Cagetti and Mariacristina De Nardi

WP-05-09

Wealth inequality: data and models
Marco Cagetti and Mariacristina De Nardi

WP-05-10

What Determines Bilateral Trade Flows?
Marianne Baxter and Michael A. Kouparitsas

WP-05-11

Intergenerational Economic Mobility in the U.S., 1940 to 2000
Daniel Aaronson and Bhashkar Mazumder

WP-05-12

Differential Mortality, Uncertain Medical Expenses, and the Saving of Elderly Singles
Mariacristina De Nardi, Eric French, and John Bailey Jones

WP-05-13

Fixed Term Employment Contracts in an Equilibrium Search Model
Fernando Alvarez and Marcelo Veracierto

WP-05-14

Causality, Causality, Causality: The View of Education Inputs and Outputs from Economics
Lisa Barrow and Cecilia Elena Rouse

WP-05-15

7

Working Paper Series (continued)
Competition in Large Markets
Jeffrey R. Campbell

WP-05-16

Why Do Firms Go Public? Evidence from the Banking Industry
Richard J. Rosen, Scott B. Smart and Chad J. Zutter

WP-05-17

Clustering of Auto Supplier Plants in the U.S.: GMM Spatial Logit for Large Samples
Thomas Klier and Daniel P. McMillen

WP-05-18

Why are Immigrants’ Incarceration Rates So Low?
Evidence on Selective Immigration, Deterrence, and Deportation
Kristin F. Butcher and Anne Morrison Piehl

WP-05-19

The Incidence of Inflation: Inflation Experiences by Demographic Group: 1981-2004
Leslie McGranahan and Anna Paulson

WP-05-20

Universal Access, Cost Recovery, and Payment Services
Sujit Chakravorti, Jeffery W. Gunther, and Robert R. Moore

WP-05-21

Supplier Switching and Outsourcing
Yukako Ono and Victor Stango

WP-05-22

Do Enclaves Matter in Immigrants’ Self-Employment Decision?
Maude Toussaint-Comeau

WP-05-23

The Changing Pattern of Wage Growth for Low Skilled Workers
Eric French, Bhashkar Mazumder and Christopher Taber

WP-05-24

8