Job Loss: Causes and Policy Implications

^Pr^ep^ar^{ed f}^o^r^M^e^m^b^ers^an^d^Com^mⁱ^t^t^ees^of^C^ong^r^e^ss

Total nonfarm private employment has fallen since the beginning of 2008. Job loss is one of the
most important macroeconomic problems facing policymakers, both in terms of its economic and
social cost. But what is often missing from the policy debate is a distinction between net job loss
and gross job loss. Gross job loss is the total number of jobs eliminated by all contracting firms in
a given period, whereas net job loss is the result of greater gross job loss than gross job gains in a
given period. Economists view net job loss as a detrimental phenomenon, and most recommend
that fiscal and monetary policy be used to mitigate it. However, they view gross job loss, as long
as it is offset by gross job gains, as a healthy and normal part of a functioning market economy,
although it may have social costs and will not affect all regions or industries equally.
Data reveal that gross job loss and job gains are each, on average, 20 times higher than net job
loss (or gains) in any given quarter. This is true in both expansions and recessions. Clearly, gross
job loss is not incompatible with a healthy labor market: during the 1990s expansion in which the
unemployment rate was lower than it had been in three decades, gross job losses steadily
increased as the expansion progressed. Even during the 2001 recession and subsequent “jobless
recovery,” gross job gains continued to average about 8 million per quarter; but these gross job
gains were more than offset by gross job losses. In the subsequent expansion, gross job gains
stayed relatively constant, but gross job losses fell. Small businesses have both higher gross job
gains and losses than large firms, and have tended to contribute modestly more net job creation.
Many causes of job loss have been offered, including imports, trade deficits, offshore
outsourcing, direct investment abroad, and restructuring. But economic theory suggests that all of
these cause gross job loss, not net job loss. Historical experience is supportive: neither imports,
the trade deficit, nor the implementation of trade liberalization agreements are correlated with net
job loss. Theory suggests, and empirical evidence has confirmed, that only recessions cause net
job loss.
Policies that impede gross job loss may seem to be a desirable way to limit net job loss at first
blush. However, such policies could make firms reluctant to hire new workers, because a firm
would not be able to subsequently reduce its workforce easily if the need for the new workers
proved to be only temporary. As a result, gross job gains could decline; if gross job gains declined
by more than gross job loss declined, net job creation would decline. International comparison
confirms this view: Germany, France, Italy, and Spain all had high barriers to job loss and
unemployment rates that were typically twice as high in the 1990s as low barrier countries like
the United States. Although attempts to impede gross job loss may reduce economic efficiency,
policy can (and does) assist some of those affected by gross job loss through unemployment
insurance and other parts of the social safety net. Whether the existing social safety net is
adequate as gross job loss increases is the subject of policy debate. This report will be updated as
events warrant.

Examining the Data.........................................................................................................................1
Displaced Workers....................................................................................................................4
Mass Layoffs.............................................................................................................................4
What Causes Gross Job Loss and Net Job Loss?............................................................................5
Tr ade .......................................................................................................................... ............... 5
Trade Deficits............................................................................................................................8
Offshore Outsourcing................................................................................................................8
Overseas Investment.................................................................................................................9
Economic Restructuring..........................................................................................................10
Economic Downturns...............................................................................................................11
Labor Market Conditions........................................................................................................12
Policy Implications........................................................................................................................13
Conclusion ..................................................................................................................................... 17
Figure 1. Gross Job Gains, Gross Job Loss, and Net Job Gains/Loss, 1992-2007..........................3
Figure 2. Effects of Trade on Employment.....................................................................................6
Figure 3. Effects of Recessions on the Labor Market, 1946-2007................................................12
Table 1. Historical Correlation Between Trade and Employment...................................................7
Table 2. International Comparison of Employment Protection Policies in 2003........................14
Author Contact Information..........................................................................................................18

otal private nonfarm employment fell from a peak of 111.6 million in February 2001 to a ¹
trough of 108.4 million in July 2003. It then expanded through 2007, reducing the
unemployment rate to a relatively low level, although not as low as was reached at the end T

of the previous expansion. Since the beginning of 2008, employment has fallen again and the
unemployment rate has risen.
Job loss—declines in employment—is one of the most important macroeconomic problems
facing policymakers, both in terms of its economic cost and the social toll it takes on our society.
But what is often missing from the policy debate is a distinction between net job loss and gross
job loss. Gross job loss is the total number of jobs eliminated by all contracting firms in a given
period, whereas net job loss is the result of greater gross job loss than gross job gains in a given
period. In expansions, the labor market is characterized by net job creation amidst gross job loss.
This is required to maintain steady employment rates with a growing population. It is only during
recessions that the overall labor market experiences persistent net job loss. Economists view net
job loss as a detrimental phenomenon and most recommend that policy be used to mitigate it.
However, they view gross job loss, as long as it is offset by gross job gains, as a healthy and
normal part of a functioning market economy, although it may have social costs.

A quarterly data series from the U.S. Bureau of Labor Statistics (BLS), shown in Figure 1, ²
provides data that help to put the distinction between gross and net job loss into perspective.
These data are measured from the firm’s perspective—changes in the size of the firm’s
workforce—not the employee’s perspective. For this reason, the data, in a sense, undercount the
amount of change in the workforce because they do not account for movements of individual
workers to and from any given firm if the firm remains the same size (e.g., a worker quits and is ³
quickly replaced by a new hire). The gross job loss figures, from the employee’s perspective, ⁴
could be involuntary (layoff, firing) or voluntary (quitting, retirement).
In the third quarter of 2007, gross job loss and gains equaled 7.5 million and 7.2 million per
quarter, respectively, each about 5% of total employment. As can be seen in Figure 1, gross job
loss and job gains are each, on average, around 20 times higher than net job loss (or gains) in any
given quarter. This is true in both expansions and recessions. The rate of job gains increased
steadily from the beginning of the series in 1992 until the end of 1999; at the same time, job

¹
^T^o^ta^{l e}^m^plo^y^m^e^{nt is}^m^e^a^s^ur^e^d^{in tw}^{o dif}^f^e^r^eⁿ^{t B}^L^{S s}^u^r^v^ey^s^,^the^hous^e^h^{old s}^u^r^v^ey^a^{nd the}^e^s^ta^blis^h^m^eⁿ^{t s}^u^r^v^ey^.
^D^a^ta^{on t}^o^ta^{l e}^m^ploy^m^e^{nt c}^ite^{d in}^this^r^e^p^o^r^t^c^o^m^e^f^r^o^m^the^e^s^ta^blis^hm^eⁿ^{t s}^u^r^v^ey^,^w^h^ic^{h que}^rⁱ^e^s^e^m^ploy^e^r^s^a^{bout t}^h^e
^num^be^r^of^jobs^on^theⁱ^r^pa^y^r^olls^{, unle}^s^s^ot^he^r^w^is^e^note^d^.
²^F^o^{r m}^o^{re i}ⁿ^f^o^rm^atⁱ^oⁿ^,^{see T}ⁱ^m^o^t^h^y^Pⁱ^vet^z^,^Mⁱ^ch^ael^S^earso^n,^an^{d J}^a^m^e^{s S}^p^l^e^t^zer,^“M^easu^rⁱⁿ^{g Jo}^b^and^E^s^t^a^b^lⁱ^s^h^m^en^t
^Flow^s^w^{ith B}^L^{S L}^ong^itudina^l^Mi^c^r^oda^ta^,”^Moⁿ^t^h^l^{y L}^a^bo^{r R}^evi^ew^{, A}^p^r^il²⁰⁰^{1, p. 13}^.
³^{Job f}^l^ow^{s f}^r^o^m^the^e^m^ploy^e^e^{s’ pe}^rspe^c^tiv^e^w^e^re^e^s^ti^m^a^te^{d in Bruc}^e^Fa^llic^k^a^{nd Cha}^r^le^{s Fle}ⁱ^sc^hm^aⁿ^{, “}^E^m^p^lo^y^e^r-to-
^Em^plo^y^e^r^Flow^s^{in the}^U^.^S.^L^a^bo^r^Ma^r^k^e^t^,”^Fe^de^r^a^{l R}^e^s^e^r^v^e^B^o^a^r^d^of^G^ove^r^nor^s^,^Fin^anc^e^an^{d Ec}^oⁿ^o^m^ic^s^Dⁱ^s^c^us^sⁱ^on
^Se^rⁱ^e^s²⁰⁰^4-³⁴^,^Ma^y^2004.^T^h^e^y^e^s^ti^m^a^te^tha^t^aⁿ^a^v^e^r^a^g^e^of^2.6%^of^em^ploy^e^e^s^c^h^aⁿ^g^e^jobs^each^moⁿ^t^h^.
⁴^D^a^ta^{on la}^y^o^f^f^s^/^dis^c^ha^r^g^es^a^{nd q}^u^its^c^aⁿ^be^f^oun^{d in}^a^s^e^pa^r^a^te^{, no}^nc^om^pa^r^a^ble^da^ta^s^e^rⁱ^e^s^f^r^o^m^B^L^{S w}^h^ic^{h w}^a^s
^st^art^e^dⁱⁿ^Decem^b^e^r²⁰⁰⁰^{. I}ⁿ^the^r^e^c^e^s^sⁱ^ona^r^y^y^e^a^r^of²⁰⁰¹^{, la}^y^o^ff^s^/^dis^c^ha^rg^e^s^a^v^e^r^a^g^e^d^{20 m}^illion^p^e^r^y^e^a^r^a^{nd quits}
^a^v^e^r^a^g^e^d^{31 m}^illion^pe^{r y}^e^a^r^{. In 2}⁰⁰⁶^{, la}^y^o^ff^s/disc^ha^rge^s^a^v^e^r^a^g^e^d^{19 m}^illio^{n pe}^{r y}^e^a^r^a^{nd qu}^{its a}^v^e^r^a^g^e^d^{32 m}^illion
^p^e^{r y}^e^ar.^T^h^{ese d}^a^t^a^{are co}^l^l^ect^edⁱⁿ^t^h^{e Jo}^{b Op}^enⁱⁿ^{g an}^d^L^a^b^o^r^T^u^rn^o^v^{er S}^u^rvey^.

losses increased steadily from 1992 to 2001.⁵ Some, but not all, of the long run increase in gross ⁶
job gains and losses can be attributed to a growing labor force. The rest of the increase indicates
that the U.S. labor force is becoming more mobile over time, but the data do not indicate whether
this is the worker’s or the firm’s decision, or both. Most of the gross job flows occur at existing
firms, and are not due to new firms opening or old firms closing.
Clearly, gross job loss is not incompatible with a healthy labor market: during an expansion in
which the unemployment rate was lower than it had been in three decades, gross job losses
steadily increased as the expansion progressed. And even during the 2001 recession and ⁷
subsequent “jobless recovery,” gross job gains continued to average about 8 million per quarter; ⁸
but gross job gains in this period were more than offset by gross job losses. In the current
expansion, job gains and losses have been modestly lower than in the second half of the 1990s.
Although gross job gains stayed relatively constant, net employment began to rise again because
gross job losses fell. There is not yet data available on whether net job loss in 2008 has been
driven primarily by gross job gains or losses.

⁵
^A^lthoug^{h t}^h^e^g^o^v^e^rⁿ^m^e^{nt doe}^sⁿ^o^{t pr}^ov^ide^of^fⁱ^cⁱ^a^l^da^ta^{on g}^r^os^s^j^{ob l}^o^s^s^aⁿ^d^g^a^ins^be^f^o^r^e¹⁹⁹²^{, pr}ⁱ^v^a^t^e^r^e^s^e^a^r^c^h
^sug^g^e^{sts tha}^t^g^r^{oss job l}^o^{ss a}ⁿ^d^g^a^{ins on}^a^la^rg^e^sc^a^l^e^a^r^e^{not ne}^w^.^Using^g^o^v^e^rn^m^e^{nt da}^ta^{, Da}^vⁱ^{s, Ha}^ltiw^a^nge^{r, a}^nd
^Sc^huh^e^s^tim^a^t^e^d^tha^t^g^r^os^s^job^g^a^ins^a^v^e^r^a^g^e^d^5.2%^a^{nd g}^r^os^s^j^o^b^l^o^s^s^a^v^e^r^a^g^e^d^5.5%^of^to^ta^{l e}^m^ploy^m^e^{nt pe}^r^qua^r^t^e^r
^{in t}^h^e^m^a^nuf^a^c^turing^se^c^t^{or f}^r^om¹⁹⁷^{2 t}^o¹⁹^88.^Ste^v^eⁿ^Da^vⁱ^{s, Joh}ⁿ^Ha^ltiw^aⁿ^g^e^{r, a}^{nd Sc}^{ott Sc}^hu^h,^J^o^{b C}^r^e^a^ti^on^an^d
^Destru^ctioⁿ⁽^C^am^b^rⁱ^d^ge,^M^A^:^M^I^{T P}^r^ess,¹⁹⁹⁶⁾^.
⁶^Fuji^ta^fⁱ^nds^tha^t^g^r^os^s^jo^{b g}^a^ins^a^{nd l}^o^s^s^e^s^due^t^o^w^o^rk^e^r^s^eⁿ^te^ring^a^{nd e}^x^iting^t^h^e^la^bor^f^o^rc^e^a^r^e^a^bout^tw^ic^e^a^s^la^r^g^e
^a^s^those^d^u^e^to^w^o^rk^e^r^{s e}ⁿ^te^ring^a^{nd e}^x^iting^une^m^p^loy^m^eⁿ^{t. (W}^orke^{rs a}^r^e^c^l^a^ssi^fⁱ^e^d^a^s^une^m^p^lo^y^e^{d if}^the^y^a^r^e^a^c^tiv^e^l^y
^se^e^k^ing^e^m^plo^y^m^e^{nt; othe}^rw^ise^,^the^y^a^r^e^c^l^a^ssi^fⁱ^e^d^a^s^out^of^the^la^b^o^{r f}^o^rc^e^.^{) Shig}^e^r^{u F}^u^jita^,^“^W^ha^{t Do W}^o^rk^e^r^Flow^s
^T^e^{ll U}^s^A^{bout C}^y^c^lic^a^l^Fluc^tua^tio^nsⁱⁿ^Em^ploy^m^e^nt,”^Fe^de^r^a^{l R}^e^s^e^r^v^e^B^a^nk^of^P^h^ila^de^l^p^hⁱ^a^,^Bu^siⁿ^e^{ss Review}^{, 200}^7:^2,
^{p. 1}^.
⁷^For^m^o^r^e^inf^o^r^m^a^{tion, s}^e^e^R^L³²⁰⁴⁷^,^T^h^{e "}^J^ob^l^e^{ss R}^eco^very"^F^r^{om t}^h^{e 20}⁰¹^R^ecessi^oⁿ^:^A^Co^m^p^a^rⁱ^s^oⁿ^t^o^E^a^rlⁱ^e^r
^Re^c^o^v^e^r^ie^s^{and P}^o^s^sⁱ^ble^Ex^p^l^a^nat^ions^,^by^Ma^r^c^L^a^bonte^a^{nd L}ⁱ^nda^L^e^vⁱ^ne^.
⁸^R^e^s^e^a^r^c^h^e^r^s^dis^a^g^r^ee^a^{bout w}^h^e^t^he^r^c^h^aⁿ^g^e^sⁱⁿ^ne^{t e}^m^ploy^m^e^{nt a}^r^e^dr^iv^eⁿ^by^c^h^aⁿ^g^e^s^{in g}^r^os^s^job^g^a^ins^or^g^r^os^s^j^ob
^losse^{s. Fuji}^ta^(o^{p. c}^it.)^fⁱ^{nds tha}^t^g^r^os^s^jo^{b los}^s^e^s^inf^l^ue^nc^e^c^y^c^lic^a^l^em^ploy^m^e^{nt pa}^tte^rⁿ^s^m^o^r^e^thaⁿ^g^r^os^s^jo^{b g}^a^ins^.
^Shim^e^r^a^{nd H}^a^{ll f}ⁱ^{nd t}^h^a^t^g^r^os^s^jo^{b g}^a^ins^d^o^mⁱ^na^te^c^h^aⁿ^g^e^s^{in ne}^{t e}^m^plo^y^m^e^{nt. Se}^e^R^obe^r^t^H^a^{ll, “}^J^o^b^L^o^s^s^,^J^o^b
^Find^ing^,^aⁿ^{d U}ⁿ^e^m^ploy^m^e^{nt in th}^e^U^.^{S. Ec}^o^nom^y^O^v^e^r^the^P^a^s^t^Fif^t^y^Ye^a^r^s^,^”^N^a^tiona^{l B}^u^r^e^a^u^of^Ec^onom^ic^R^e^s^e^a^r^c^h^,
^w^o^r^k^ing^pa^pe^r¹¹⁶⁷⁸^{, O}^c^to^be^r²⁰^{05; R}^obe^r^t^Shim^e^r^{, “}^R^e^a^s^s^e^ss^ing^the^Iⁿ^s^aⁿ^{d O}^u^ts^of^Uⁿ^e^m^ploy^m^e^nt,”^w^o^r^k^ing
^pa^pe^r^,^A^u^g^u^s^t²⁰⁰⁷^.

Figure 1. Gross Job Gains, Gross Job Loss, and Net Job Gains/Loss, 1992-2007
10000
8000
6000nds
a
4000s
2000Thou
0
-2000
2 993 994 995 996 997 998 999 0 1 2 3 4 5 6 7
1 99 1 1 1 1 1 1 1 200 200 200 200 200 200 200 200
GainsLossesNet
^Source:^BLS.
It is often claimed that small businesses are the engine of job creation in the U.S. economy. To an
extent, this is a misconception based on the confusion between net and gross job flows. Firms
with 99 employees or fewer, which account for 38.1% of total private employment, accounted for
61.1% of gross job gains between 1992:3 and 2005:1. But while these firms had a
disproportionate share of gross job gains, they also had a disproportionate share of gross job loss,

62.1% of the total. On net, they accounted for 46.3% of net job gains over that time period—

modestly more than their share of total employment, but significantly less than gross flows would
indicate. (Recent employment trends followed a similar pattern.) Interestingly, in the 2001
recession and jobless recovery, very large firms accounted for a disproportionate share of net job ⁹
loss.
Overall, these data provide a picture, during expansions, of a highly dynamic U.S. labor market in
which labor rapidly shifts from firm to firm to its most efficient use. This vitality is the essence of ¹⁰
economic growth and rising living standards for society as a whole in a market economy. It is
caused both by output shifting from some firms to more efficient ones within an industry and by
shifts in spending from one industry to another, due to factors such as changing consumer tastes,
technology, or comparative advantage. Of course, there will always be winners and losers in a
market economy. Although significant gross job loss is consistent with net job creation (because it
is offset by gross job gains) for the nation as a whole, gross job loss can translate into net job loss

⁹
^A^{ll da}^ta^f^r^o^m^B^u^r^e^a^u^of^L^a^bor^Sta^tis^tic^s^,^“^N^ew^Q^u^a^r^te^r^l^y^D^a^ta^f^r^om^B^L^{S on B}^u^sⁱ^ne^s^s^Em^ploy^m^e^{nt Dy}^namⁱ^c^s^b^y
^Sⁱ^{ze o}^f^Fⁱ^rm^,^”ⁿ^e^w^s^rel^e^a^s^{e US}^D^L⁰⁵^-²²⁷⁷^,^Decem^b^e^{r 8}^,²⁰⁰⁵^.
¹⁰^T^o^judg^e^the^be^ne^f^its^of^g^r^os^s^jo^{b los}^s^f^r^om^the^pe^r^s^pe^c^tiv^e^of^a^na^tion’^s^e^c^o^nom^ic^de^v^e^lopm^eⁿ^{t, c}^ons^ide^r^the
^c^ounte^r^e^x^a^m^ple^:^if^the^U^.^{S. ha}^{d e}^x^pe^rⁱ^eⁿ^c^e^d^{no g}^r^os^s^jo^{b los}^sⁱⁿ^t^h^e²⁰^th^c^eⁿ^tur^y^{. A}^t^the^be^gⁱ^nning^o^f^the²⁰^th^cen^t^u^ry^,
^m^u^c^h^of^the^U^.^{S. p}^o^p^u^la^ti^{on w}^o^r^k^e^dⁱⁿ^the^f^a^r^m^ing^s^e^c^t^or^{. I}^f^g^r^os^s^{job l}^o^s^sⁱⁿ^the^f^a^r^m^ing^s^e^c^t^or^ha^d^not^oc^c^u^r^r^e^d,
^thos^e^w^o^r^k^e^r^s^w^ould^not^ha^v^e^be^eⁿ^a^v^a^ila^ble^{to w}^o^r^k^{in t}^h^e^g^r^ow^ing^m^a^nuf^a^c^t^ur^ing^a^{nd s}^e^r^v^ic^e^s^e^c^t^or^{. T}^oda^y^,^f^a^r^m^s
^a^c^c^{ount f}^o^r^a^b^out^1.^5%^of^the^to^ta^{l e}^m^plo^y^m^e^{nt a}^{nd 1%}^of^the^na^tio^n’^s^o^u^tp^{ut. I}^f^g^r^os^s^jo^{b l}^o^s^s^ha^dⁿ^o^{t oc}^c^u^r^r^e^{d in}^the
^la^s^t^c^eⁿ^tur^y^,^m^u^c^h^of^toda^y^’^s^la^bor^f^o^r^c^e^w^ould^not^be^a^v^a^ila^ble^to^pr^o^duc^e^the^othe^r^99%^of^U^.^{S. o}^u^t^put^.

at the local level even when national employment is rising because the losses and gains may not
occur in the same geographic area.
Furthermore, while steady net employment gains are unambiguously good for society as a whole,
the data do not necessarily indicate that the same individuals who lose jobs also gain jobs. The
data also do not indicate whether the job loss is voluntary or involuntary, nor how many of the
individuals who involuntarily changed jobs were forced to take new jobs that were less desirable
or paid less. A separate (and noncomparable) data series on worker displacement from BLS can ¹¹
help to answer these questions. BLS classifies workers as displaced if they lost their job because
their plant closed down or moved, their positions or shifts were abolished, or there was
insufficient work. From 2003 to 2005, 3.8 million workers with tenure of three years or more
were displaced (another 4.3 million short-tenured workers were displaced during that period).
Although the two data series cannot be compared directly, gross job loss equaled 90 million over
that three-year period. Displacement is significantly higher during recessions; for example, from

2001 to 2003, 5.3 million long-tenured workers were displaced.

Of the displaced workers, about 70% were reemployed, 13% were unemployed, and 17% had left
the labor force at the beginning of 2004. Of those reemployed full-time, about 51% were now
earning more than they had at the displaced job, and 29% were now earning significantly lower
wages (at least 20% lower). Workers 55 years of age and older had lower reemployment rates ¹²
than younger workers. Displaced workers fared better during expansions. For example, in 2001-
2003, 20% of displaced workers were unemployed, and 33% of those re-employed had
significantly lower wages.
Some gross job loss takes the form of mass layoffs, during both expansions and recessions. In
another (non-comparable) survey from BLS, 0.9 million workers lost their jobs from extended
mass layoffs in the four quarters ending 2008:1. This figure undercounts workers affected by
mass layoffs because it does not include mass layoffs of less than 50 workers or layoffs that lasted
less than 30 days. Mass layoffs tend to be cyclical: workers separated by mass layoffs rose from
1.2 million in 2000 to 1.5 million in 2002. Since unemployment totaled 8.4 million in 2002, mass
layoffs are an important but not primary cause of unemployment. BLS has not kept a continuous
data series long enough to determine if there has been a long-term upward trend in mass layoffs ¹³
beyond the cyclical trend.

¹¹
^B^L^{S, “}^W^or^k^e^r^Dⁱ^s^p^la^c^e^m^e^{nt, 20}^03-²⁰^05,”^pr^e^s^s^r^e^le^a^s^e^U^S^D^L^06-¹⁴^54,^A^u^g^u^s^t^17,²⁰⁰⁶^{. T}^h^e^ne^w^e^s^t^a^v^a^ila^ble^da^ta
^{in t}^h^is^s^e^rⁱ^e^s^a^r^e^f^r^o^m²⁰⁰⁵^{. A}^{ll o}^f^the^f^o^llowⁱ^ng^da^ta^r^e^f^e^r^{to w}^o^r^k^e^r^s^w^{ith thr}^e^e^y^e^a^r^s^of^te^nur^e^or^m^o^r^e^{. Se}^e^a^l^s^o^C^R^S
^R^e^por^{t R}^L³²²⁹^2,^O^ffs^horⁱ^{ng (}^a^.k^.^a^{. O}^ffs^h^o^r^e^O^u^ts^our^cⁱⁿ^g⁾^aⁿ^d^J^o^b^Iⁿ^s^e^c^u^r^ity^Am^o^{ng U}^.^{S. W}^o^r^k^e^r^s^{, by}^Lⁱ^nda^L^e^vⁱ^ne^.
¹²^F^a^rb^{er f}ⁱⁿ^d^s^t^h^at^o^l^d^e^{r di}^sp^l^aced^w^o^rker^s^w^e^re^m^o^{re l}ⁱ^k^el^y^t^o^su^ff^e^r^{a si}^gnⁱ^fⁱ^can^t^l^o^{ss o}^f^earnⁱⁿ^g^{s t}^h^aⁿ^y^oung^e^r
^w^o^rk^er^{s. Hen}^r^y^Farb^er,^J^ob^Los^sⁱⁿ^the^Uⁿ^ite^d^St^ate^s^{, 1}⁹⁸¹^-²⁰⁰¹^{, N}^a^tiona^{l B}^u^r^e^a^u^of^Ec^on^om^ic^R^e^s^e^a^r^c^h^{, W}^o^r^k^ing
^P^a^pe^r⁹⁷⁰⁷^{, Ma}^y²⁰⁰^3.
¹³^For^m^o^r^e^inf^o^r^m^a^{tion, s}^e^e^C^R^S^R^e^por^{t R}^L³⁰⁷⁹^9,^Uⁿ^e^m^ploy^m^e^nt^Thr^o^ug^{h L}^a^y^o^ffs^an^{d O}^ffs^h^o^r^e^O^u^ts^our^cⁱⁿ^g^{, by}
^Lⁱ^nda^L^e^vⁱ^ne^.

Several economic phenomena have been identified in popular discussion as purportedly causing
job loss. Although all of these phenomena cause gross job loss, most have a much smaller effect
on net job loss than popularly perceived. The exception is the business cycle: in each instance,
recessions have been the cause of persistent net job loss in the post-war period.
When trade expands, greater imports cause gross job loss, as products that were previously
produced in the United States are now produced by workers in other countries, rendering those
U.S. workers redundant. However, economic theory states that expansions in trade have no effect
on net employment. As foreign countries increasingly exchange their goods for U.S. exports,
more workers are needed in U.S. export industries. In addition, because trade is based on
comparative advantage, trade increases the purchasing power of U.S. incomes in the aggregate.
Thus, trade allows the U.S. economy as a whole to produce and consume more domestic goods, ¹⁴
requiring more workers to produce them. It is possible that there could be some transitional loss
in net employment if workers cannot easily be reallocated into other sectors of the economy,
causing net employment to temporarily be greater than zero. For example, workers who have lost
their jobs in the import-competing industries may not have the skills needed by export industries. ¹⁵
But this transitional effect would disappear once markets had adjusted.
U.S. history offers persuasive evidence that trade liberalization has no effect on net employment,
as can be seen in Figure 3. During the post-war period, U.S. trade has become progressively
liberalized, with eight rounds of world trade liberalization negotiated between 1947 and 1993
through the General Agreements on Tariffs and Trade (GATT, later became World Trade
Organization), as well as the Canadian Free Trade Agreement in 1989 and North American Free
Trade Agreement (NAFTA) in 1994. Imports have increased steadily as a percentage of GDP
throughout the post-war period, from about 4% of GDP in the 1940s to about 14% of GDP in
recent years. If trade caused net job loss, employment would have declined and unemployment
risen throughout the post-war period. The opposite is the case: employment has steadily increased
during the post-war period, and the unemployment rate has mirrored the business cycle, not trade
patterns. Indeed, trade liberalization does not appear to have strong effects on even transitional
unemployment. For example, NAFTA was implemented when aggregate employment was rising
and unemployment was falling. GATT Rounds 1, 3, 4, 6, 7, and 8 were completed when
unemployment was low, and unemployment, though high, fell subsequent to GATT Rounds 2 and
5. The most recent example the United States has with significantly increasing trade restrictions

¹⁴
^For^m^o^r^e^inf^o^r^m^a^{tion, s}^e^e^C^R^S^R^e^por^{t R}^L³²⁰⁵^9,^Tr^ade^,^Tr^a^d^e^B^a^r^rⁱ^e^r^s^,^an^{d Tr}^a^d^e^D^e^fic^its^:^I^m^plⁱ^c^atio^ns^for^U^.^S.
^Ec^onom^ic^W^e^lfar^e^,^by^Craⁱ^g^K^.^Elw^e^ll.
¹⁵^E^v^enⁱ^f^U.^S^.^w^o^{rkers co}^u^l^d^no^t^c^o^m^p^et^e^wⁱ^t^h^f^o^rei^gⁿ^w^o^rker^s,^t^h^e^f^l^o^a^tⁱⁿ^{g exch}^an^{ge rat}^e^s^y^st^e^m^w^o^u^l^d^p^r^even^t
^im^{ports f}^r^o^m^le^a^d^ing^{to ne}^t^jo^{b l}^o^{ss. Sinc}^e^d^o^lla^{rs m}^u^{st be}^so^l^dⁱⁿ^t^h^{e f}^o^rei^gⁿ^exch^an^ge^m^a^r^k^et^t^o^p^u^rch^a^{se i}^m^p^o^r^t^s^,ⁱ^f
^f^o^re^ig^{n f}ⁱ^r^m^s^be^c^a^m^e^m^o^re^“c^om^pe^titiv^e^”^a^t^pro^ducⁱⁿ^g^g^oods^t^h^aⁿ^U^.^{S. f}ⁱ^rm^s^s^u^c^h^tha^t^im^ports^ros^e^,^the^d^o^lla^r
^e^x^c^h^aⁿ^g^e^ra^te^w^{ould de}^pre^cⁱ^a^t^e^t^h^ro^ug^{h the}^purc^h^a^s^e^of^im^{ports until U.}^{S. f}ⁱ^rm^s^w^e^r^e^“^c^o^m^p^e^titiv^e^”^a^g^aⁱ^n.

was the Smoot-Hawley tariffs, which did not stem the loss of employment during the Great ¹⁶
Depression.
Figure 2. Effects of Trade on Employment
^Source:^Burea^{u of}^{Economic An}^{alysis, Bureau of Labor Statistics.}
^Note:^{Import and Trade Balance}^{data expr}^{essed in real}²⁰^{00 dollars. The first seve}^{n ro}^{unds of}^GA^TT
^{negotiations are labeled GATT 1-7; the eight}^{h rou}^{nd is labeled the}^Uruguay^{Round. The Canadian F}^{ree Trade}
^{Agreement is labeled CFTA.}
Although regression analysis, which allows other factors to be held constant, is beyond the scope
of this report, informal quantitative evidence on the relationships portrayed in Figure 3 can be ¹⁷
gleaned using correlation analysis. The results are presented in Table 1, which shows that

¹⁶
^Using^plaⁿ^t-^le^v^e^{l da}^ta^{, Da}^vⁱ^{s e}^t^a^l^{. de}^m^onstra^t^e^tha^t^tra^d^e^ha^{s litt}^le^e^ffe^c^t^on^gr^os^s^jo^{b l}^o^{ss e}^ithe^r^{. A}^lthoug^h^the^y
^f^{ound t}^h^a^tⁱⁿ^d^u^s^t^rⁱ^e^s^w^{ith the}^hig^h^e^s^{t de}^g^r^e^e^ofⁱ^m^por^{t pe}^ne^tr^a^tion^ha^{d t}^h^e^hig^h^e^s^{t g}^r^os^s^j^ob^los^s^r^a^te^s^,^the^y^ha^{d the}
^s^e^c^{ond h}ⁱ^g^h^e^s^{t g}^r^os^s^jo^{b g}^a^{in r}^a^te^{, a}^{nd f}ⁱ^r^m^s^w^{ith the}^s^e^c^oⁿ^d^hⁱ^g^h^e^s^{t de}^g^r^e^e^ofⁱ^m^por^{t pe}^ne^tr^a^tⁱ^on^ha^{d the}^low^e^s^t^g^r^os^s
^job^{loss ra}^te^{s. A}^f^te^{r c}^ontrolling^f^o^{r w}^a^g^e^{s (a}^{s a}^proxy^f^o^{r c}^o^m^p^a^r^a^tⁱ^v^e^a^d^v^a^nta^g^e⁾^{, the}^re^la^tions^hip^be^tw^e^e^{n g}^r^{oss job}
^l^o^{ss an}^dⁱ^m^p^o^r^t^p^eⁿ^e^t^r^atⁱ^o^{n b}^{ecame st}^atⁱ^s^tⁱ^cal^l^yⁱⁿ^si^gnⁱ^fⁱ^can^t^.^{As t}^h^{e au}^t^h^o^r^{s po}ⁱⁿ^t^ou^t^,^t^h^eo^ry^su^gge^st^{s t}^h^at^t^h^{e ef}^f^ect^o^f
^tr^a^d^e^libe^r^a^liz^a^tio^{n o}ⁿ^g^r^os^s^j^{ob f}^l^ow^s^is^am^big^uous^{. L}ⁱ^be^r^a^liz^a^tion^c^ould^inc^r^e^a^s^e^g^r^os^s^{job f}^l^ow^s^by^e^x^pos^ing^the^U^.^S^.
^e^c^onom^y^{to m}^o^r^e^s^hoc^k^s^f^r^om^a^b^r^o^a^d^or^de^c^r^e^a^s^e^g^r^os^s^{job f}^l^ow^s^by^div^e^r^sⁱ^f^yⁱ^ng^the^e^c^onom^y^s^o^tha^t^the^e^f^f^e^c^t^s^of
^dom^e^s^tic^s^hoc^k^s^a^r^e^re^duc^e^d^{. Ste}^v^eⁿ^D^a^vⁱ^s^,^J^{ohn H}^a^ltiw^aⁿ^g^e^{r, a}ⁿ^d^Sc^ott^Sc^hu^h,^Job^C^r^ea^ti^on^an^{d Dest}^ru^ctⁱ^oⁿ
⁽^C^a^m^b^rⁱ^d^ge,^M^A^:^M^I^{T P}^r^ess,¹⁹⁹⁶⁾^.
¹⁷^Co^rrel^a^tⁱ^oⁿⁱ^s^m^easu^r^ed^oⁿ^{a scal}^{e o}^f¹^t^o^-¹^,^w^h^{ere co}^rrel^a^tⁱ^oⁿ⁼¹ⁱ^f^t^w^o^vari^ab^l^e^{s are p}^e^r^f^ect^l^y^co^rre^l^a^t^e^d^(al^w^a^y^s
⁽^c^onti^nue^d.^..)

between 1946 and 2007, changes in employment are highly correlated with changes in imports.
Thus, the historical experience is the opposite of the typically claimed relationship: when imports
increased, employment typically also increased. Although this is not evidence that higher imports
cause higher employment—the two variables are correlated because both usually increase—it is
evidence that higher imports do not cause lower employment. The table also demonstrates that
the implementation of trade liberalization agreements has virtually no relationship historically to
changes in net employment in the same or following year (in case there is a lagged effect), as
economic theory would suggest. In sum, the results suggest that trade either has no negative
effect on employment, or the effect is swamped by other factors.
Table 1. Historical Correlation Between Trade and Employment
^{Correlation between employment growth}^{and ...}^{Correlation C}^oefficient
^{... change in imports (as a %}^{of G}^DP)^0.40
^{... maj}^{or trade agreement in}^same^{or previous yea}^r^0.12
^{... change in trade balance}^{(as a %}^{of GDP)}^-0.01
^{... GDP growth}^0.52
^Source:^{CRS calculations based}^{on BLS, BEA data}
^Note:^Correlat^{ion based on an}ⁿ^{ual data 19}^46-2⁰⁰^{7; major trade}^{agreements i}^{nclude GATT rou}^{nds 1 to 8,}
^{Canadian Free Trade Agreeme}^nt^{and North A}^{merican Fre}^{e Trade}^Agreement.
Survey data from mass layoffs does not identify trade as a major source of gross job loss either.
For example, only 2,900 of the 301,400 workers laid off in the first quarter of 2008 reported
import competition to be the cause of the layoff.
A recent study found that trade had a limited effect on net job loss in the recent recession and ¹⁸
jobless recovery. It found that the industries with the greatest job loss during that period
included both those affected (business services, manufacturing) and unaffected (leisure and
hospitality, transportation, construction, and communications) by trade and outsourcing. It then
measured the number of American workers that would be needed to produce U.S. imports
compared to the number of workers that are needed to produce U.S. exports, and found that the
difference amounted to only 2.4% of total employment in 2003. This estimate should not be
interpreted as how much employment would rise in the absence of trade since workers affected
by trade may be re-employed producing non-tradable goods.
Some policymakers are particularly concerned that trade is responsible for the continuing decline
in manufacturing employment in recent years. Even after employment began increasing in the rest
of the economy in 2003, manufacturing employment has continued to fall, from 17.6 million in
1998 to 13.9 million in 2007. Yet trade cannot be the primary cause of this decline because

^(...c^on^tin^ue^d)
^m^ove^tog^e^the^{r), c}^o^rre^la^tioⁿ⁼^{0 if}^t^w^{o v}^a^ria^b^le^{s a}^r^e^{not c}^o^rre^la^te^{d (d}^{o n}^o^{t m}^o^v^e^tog^e^the^{r), a}^{nd c}^o^rre^la^tion=^-1^if^the^tw^o
^v^a^ria^b^le^{s a}^r^e^pe^r^f^e^c^t^l^y^ne^ga^tiv^e^l^y^c^o^rre^la^te^{d (a}^lw^a^y^s^m^o^v^e^{in opp}^os^ite^dire^c^tio^ns).
¹⁸^Er^ic^a^G^r^os^he^{n, e}^t^a^l^{, “}^U^.S.^J^o^bs^G^a^ine^d^a^{nd L}^o^s^t^T^h^r^oug^{h T}^r^a^d^e^:^A^N^e^{t Me}^a^s^ur^e^,^”^Fe^de^r^a^{l R}^e^s^e^r^v^{e B}^a^nk^of^N^e^w
^Yo^rk,^C^u^r^r^e^{nt I}^s^s^u^e^sⁱⁿ^Ec^o^nomⁱ^c^s^and^Fin^anc^e^{, v}^o^l^.¹¹^,^no^{. 8}^,^A^u^g^u^s^t²⁰^05.

manufacturing output has grown by 22% in real terms over those years. By identity, employment
can fall as output rises only if productivity is rising faster than output. So the decline in
manufacturing employment must be primarily attributable to rapid technological change and ¹⁹
efficiency gains, not trade.
Some who concede that trade has no effect on net employment when higher imports are matched
by higher exports argue that trade nevertheless reduces net employment when higher imports are
matched instead by a larger trade deficit. They reason that higher imports cause gross job loss, but
are not offset by gross job gains in the export sector if they lead to a trade deficit. While this is
true, trade deficits do lead to gross job gains in other ways. When the United States runs a trade
deficit, it exchanges foreign imports for U.S. assets. This puts downward pressure on U.S. interest
rates, stimulating spending on physical investment (plant and equipment). Lower interest rates
also stimulate spending on housing and interest-sensitive goods, such as automobiles and
appliances. As a result, the trade deficit causes gross job gains in the sectors that produce plant,
equipment, housing, and interest-sensitive goods, all else equal. These gross job gains may not
occur instantaneously—so there could be transitional net job loss—but when they do occur, they
will offset the gross job loss caused by higher imports so that the trade deficit causes no net job
loss.
As can be seen in Figure 3, the historical experience confirms this conclusion: the large increase
in the trade deficit in the 1980s and 1990s took place at a time of rising employment and falling
unemployment. While the trade deficit rose during the 2001 recession and jobless recovery, it
continued to rise from 2003 to 2007 as unemployment fell. This suggests that some other factor,
such as strong aggregate demand growth, tends to simultaneously push the trade deficit up and
unemployment down. In 2008 (to date), unemployment rose despite a decline in the trade deficit.
Table 1 demonstrates that there was almost no correlation between changes in the trade balance ²⁰
and changes in employment.
The term offshore outsourcing or offshoring is frequently used in several different ways.²¹ It can
refer to U.S. multinational firms shifting production from the United States to an overseas
subsidiary, U.S. firms importing intermediate goods from foreign companies, importing services,
or U.S. firms making overseas investments. Economists view the first three phenomena similarly
to trade (the latter phenomenon will be discussed separately in the next section). When U.S. firms
outsource production to foreign firms, gross job loss occurs because goods and services that were

¹⁹
^For^m^o^r^e^inf^o^r^m^a^{tion, s}^e^e^C^R^S^R^e^por^{t R}^L³²³⁵^0,^Deⁱⁿ^dustri^a^liz^a^tion^of^the^U.^{S. Ec}^on^omy^:^The^R^o^le^{s of Tra}^d^e^,
^Pr^oduc^tⁱ^v^ity^{, a}^nd^R^ecessi^oⁿ^,^by^C^r^aⁱ^g^K^.^Elw^e^ll.
²⁰^For^m^o^r^e^inf^o^r^m^a^{tion, s}^e^e^C^R^S^R^e^por^{t R}^L³⁰⁵³^4,^Am^e^r^ic^a’^s^G^r^o^w^ing^Tr^ade^D^e^fic^it:^I^t^s^C^a^us^e^aⁿ^d^W^{hat I}^t^M^e^ans
^{for the}^Ec^on^omy^{, by}^Ma^r^c^L^a^bont^e^a^{nd G}^a^{il E. Ma}^kⁱ^ne^n.
²¹^For^m^o^r^e^inf^o^r^m^a^{tion, s}^e^e^R^L³²²⁹²^,^O^ffs^h^o^rⁱⁿ^g⁽^a^.k^.a^{. O}^ffs^h^o^r^e^O^u^ts^our^cⁱⁿ^g⁾^an^{d J}^o^{b I}ⁿ^s^e^c^u^r^ity^Am^on^{g U}^.^S^.
^Wo^rkers^{, b}^y^Lⁱ^nda^L^e^vⁱ^ne^{; a}^{nd C}^R^{S R}^e^por^t^R^L³²⁴⁸⁴^,^For^eⁱ^{gn O}^u^ts^our^cⁱⁿ^g^:^Ec^o^nom^ic^I^m^plic^ati^ons^an^{d P}^o^lic^y
^Re^s^pons^e^s^,^by^Craⁱ^g^K^.^El^w^e^ll.

being produced by U.S. workers are now being produced by workers in other nations. When
outsourcing occurs, those foreign firms must be paid using U.S. dollars. The foreign firms, in
turn, can use those dollars in three ways. First, they can buy U.S. exports, resulting in gross job
gains in the export sector. Second, they can buy U.S. assets (increase the trade deficit), resulting
in gross job gains in the interest-sensitive sectors that produce plant, equipment, housing, and
interest-sensitive goods, as explained in the previous section. Third, they can sell their U.S.
dollars for another currency, causing the dollar to depreciate; as a result, the output of U.S.
exporting firms and U.S. import-competing firms would increase as U.S. goods become more
price competitive internationally. Thus, in all three scenarios, gross job loss is offset by gross job
gains so that there is no net job loss, although there may be for a transitional period.
Quality data on outsourcing is scarce because the term has only recently been coined, there is not
yet any consensus as to how it should be defined, and the concept cannot easily be measured
accurately. It is popularly used to mean net job loss, but conceptually the term applies only to
gross job loss. A net measure of outsourcing’s effects would be hard to calculate because it would
be difficult for BLS to measure gross job gains caused by foreign firms outsourcing to the United
States, and impossible to trace the rise in employment caused by the spending on U.S. goods of
foreign firms that U.S. firms have engaged in outsourcing. Changes in imports of services
suggests that outsourcing, by that definition, is a minor phenomenon relative to total gross job
loss. The one official data source on outsourcing comes from the BLS mass layoffs series, and
this only includes one type of outsourcing, U.S. firms relocating work abroad. In the first quarter
of 2008, only 1,200 employees out of 301,400 were subject to extended mass layoff because of ²²
relocation abroad.
The argument is sometimes made that when U.S. firms decide to undertake direct investment
abroad, it reduces U.S. employment. According to the argument, if the U.S. firm had not, say,
built the new factory abroad that employs foreign workers, it would have built a factory in the
United States that employed U.S. workers. As a result, net employment is lower than it would
have been.
Economic theory states that capital investment determines the wages of employees, not the level
of employment. That is because capital investment increases the productivity of existing workers,
and in a competitive market wages are determined by productivity. If the level of employment
was based on the amount of capital available per worker, then the United States would not have
been able to achieve full employment in the past since capital per worker has increased steadily
over time. Yet throughout its history, the United States has achieved full employment most of the
time.
Even if U.S. investment abroad did lead to net job loss, the data do not suggest that this
explanation is possible since the United States is a net recipient of foreign capital, and has been
for the past few decades. In other words, any jobs hypothetically lost by U.S. investment abroad

²²
^Becau^{se t}^h^{e su}^rve^y^q^u^e^stⁱ^o^{n i}^s^new^,ⁱ^tⁱ^s^dⁱ^f^fⁱ^cu^l^t^t^o^j^u^d^g^{e i}^t^s^accu^rac^y^;^f^o^{r exa}^m^p^l^e,^t^h^{e res}^pons^e^c^oul^{d be}^bia^s^e^dⁱⁿ
^f^a^v^o^{r o}^f^o^v^er-rep^o^r^tin^g^if^o^u^t^s^o^u^r^cin^g^{is b}^eⁱⁿ^g^u^s^ed^{as a scap}^eg^o^a^{t f}^o^{r o}^t^h^e^{r reaso}ⁿ^s^f^o^{r lay}^o^ff^s.

have been more than offset by the jobs created by greater foreign investment in the United States.
Although U.S. direct investment abroad exceeds foreign direct investment in the United States,
overall foreign investment in the U.S. (direct, portfolio, and official) is greater than U.S. ²³
investment abroad. Ultimately, foreign investment has the same effect on the economy
regardless of the form it takes.
Net foreign investment in the United States is, by identity, equal to the U.S. current account
deficit (trade deficit plus net transfers and net investment income). That is because capital cannot
flow into the country on net (net borrowing cannot occur) unless the United States imports more
than it exports. Any time U.S. investment abroad increases, all else equal, the trade deficit must
fall. Thus, explanations of net job loss attributed to the trade deficit and explanations of net job
loss attributed to U.S. investment abroad are mutually exclusive—both stories cannot be correct.
This can be proven by considering the foreign exchange market. If U.S. direct investment in the
euro area increased, dollars must be exchanged for euros. This causes the dollar to depreciate
against the euro, causing U.S. exports to the euro area to rise and euro imports to the United
States to fall. As a result, the trade deficit narrows.
Declines in aggregate employment are often blamed on restructuring in the economy. For
example, the decline in employment from 2001 to 2003 is often attributed to the collapse of the
“dot-com” industry. According to this argument, resources had been overinvested in the dot-com
industry in the late 1990s. When this situation was rectified in the late 1990s, workers in that
industry were no longer needed, causing overall employment to decline.
The reallocation of resources in the economy is probably the primary reason that gross job loss
occurs. Changes in tastes, technology, and comparative advantage continually cause labor and
capital to be shifted from one industry to another in a market economy. As this report has
demonstrated, sizeable reallocations of labor across industries has been a constant in the United
States for as long as data has been collected. But in most years, the economy has not had any
problem offsetting gross job loss with a greater number of gross job gains. Thus, economic
restructuring typically is not accompanied by net job loss overall, even though it often results in
net job loss at the local level. It is possible that if restructuring were unusually large at any given
time, perhaps like the dot-com collapse, the economy could be unable to absorb that many
workers in new jobs fast enough to prevent net job loss. Unfortunately, there is no way to
systematically identify which restructuring events are large enough and separate their effects on
net job loss from other economic phenomena occurring simultaneously. For example, the net job
loss associated with the 2001 recession could be caused by nothing more than an unrelated
decline in aggregate spending, as is typical of other historical recessions. In the absence of this
decline in aggregate spending, it is possible that the dot-com collapse would not have had any
effect on aggregate employment.

²³
^P^o^rtf^o^li^{o inv}^e^stm^e^{nt re}^fe^{rs to the}^purc^h^a^s^e^of^fⁱ^na^nc^ia^{l se}^c^u^r^itie^s.^Offⁱ^cⁱ^a^l^inv^e^st^m^e^{nt re}^f^e^r^s^{to the}^p^u^rc^ha^se^of^a^sse^ts
^by^g^ove^rⁿ^m^e^{nt tr}^e^a^s^ur^ie^s^or^c^eⁿ^tr^a^l^ba^nk^s^.

Even if restructuring does not cause net job loss, there is debate as to whether recessions have the
beneficial effect of hastening restructuring. Some argue that when times are tough, firms are ²⁴
forced to innovate to survive and during booms weak firms are propped up by prosperity. If
correct, this points to an economic benefit from recessions, in contrast to the mainstream
economic view that recessions are economically wasteful because they cause productive labor
and capital to lay idle. But it is difficult to evaluate this argument quantitatively.
Unlike the other factors described above, economic downturns are the only factor that causes both
gross and net job loss according to economic theory. This theory is borne out by historical
experience, as illustrated in Table 2, which show the high correlation between contemporaneous
GDP growth and employment growth. (In fact, falling employment is part of the official
definition of a recession.) Economic downturns are characterized by insufficient aggregate
spending to support existing labor and capital resources. As a result, capital goes idle and workers ²⁵
are laid off until spending revives. The government can boost aggregate spending back to “full
employment” through expansionary fiscal policy (a larger government budget deficit) or
monetary policy (lower short term interest rates by the Federal Reserve). Net job loss caused by
economic downturns is temporary because insufficient aggregate spending is only a temporary
phenomenon—in the long run, markets adjust to bring spending back into line with potential
production. Since 1947, net job loss lasting more than one quarter has only occurred during or
immediately following a recession. There have only been two recoveries in which net job loss has
continued for more than one quarter after the recession had ended, the recoveries beginning in ²⁶

1991 and 2001.

²⁴
^By^coⁿ^t^rast^,^Cab^a^l^l^e^ro^an^{d Ham}^o^u^r^argu^{e t}^h^at^recessi^oⁿ^s^act^u^a^l^l^y^ret^a^rd^rest^ru^ct^u^rⁱⁿ^{g b}^ecau^{se h}ⁱ^st^o^rⁱ^cal^l^y^t^h^e^y^h^a^ve
^be^eⁿ^f^o^llow^e^{d by}^pe^rⁱ^ods^of^low^e^r^r^a^te^s^of^g^r^os^s^{job l}^o^s^s^,^r^a^the^r^t^h^aⁿ^hig^h^e^r^r^a^te^s^of^g^r^os^s^{job g}^a^ins^.^Rⁱ^c^h^a^r^{d C}^a^ba^lle^r^o
^a^{nd M}^oha^m^e^{d H}^a^m^our^,^T^h^e^C^o^s^t^s^{of Re}^c^e^s^sⁱ^ons^Re^vⁱ^s^ite^d:^{A Re}^v^e^r^s^e^-^L^iquid^a^ti^onis^t^Vie^w^{, N}^a^ti^ona^l^B^u^r^e^a^u^of
^Ec^onom^ic^R^e^s^e^a^r^c^h^{, W}^o^r^k^ing^P^a^pe^r⁷³⁵⁵^{, Se}^p^t^e^m^be^r¹⁹⁹⁹^.
²⁵^For^m^o^r^e^inf^o^r^m^a^{tion, s}^e^e^C^R^S^R^e^por^{t R}^L³¹²³^7,^The²⁰⁰¹^Ec^on^om^ic^Re^c^e^s^sⁱ^on:^H^o^w^Lo^ng,^H^o^w^D^e^e^p^{, an}^{d H}^o^w
^Diffe^re^{nt From t}^h^e^Past^?^,^by^Ma^r^c^L^a^bonte^a^{nd G}^a^{il E.}^Ma^kⁱ^ne^n.
²⁶^For^m^o^r^e^inf^o^r^m^a^{tion, s}^e^e^C^R^S^R^e^por^{t R}^L³²⁰⁴^7,^T^h^{e "}^J^o^b^l^{ess R}^eco^very"^F^r^o^m^t^h^{e 20}⁰¹^R^ecessi^on^{: A}^C^o^mp^a^rⁱ^s^oⁿ
^{to E}^a^r^lie^r^Re^c^o^v^e^r^ie^s^and^Pos^sⁱ^b^l^e^Ex^pla^nati^ons^,^by^Ma^r^c^L^a^bonte^a^{nd L}ⁱ^nda^L^e^vⁱ^ne^{. A}^f^te^r^s^lig^htl^y^rⁱ^sⁱ^ng^f^o^r^a^f^e^w
^m^{onths in}²⁰⁰²^{, e}^m^ploy^m^e^{nt be}^g^a^{n f}^a^lling^a^g^aⁱ^{n in 2}⁰⁰³^{. It}^{is n}^o^t^c^l^e^a^r^if^the^e^m^ploy^m^e^{nt de}^c^lineⁱⁿ²⁰⁰³^sh^oul^d^be
^c^ounte^d^a^s^pa^r^t^of^the^p^o^s^t^-^r^e^c^e^s^sⁱ^on^de^c^line^or^a^s^e^pa^r^a^te^pheⁿ^o^m^e^non^.

Figure 3. Effects of Recessions on the Labor Market, 1946-2007
^Source:^{National Burea}^{u of E}^{conomic Research; Bu}^{reau of}^Labor^Statistics.
In the long run, the effects of business cycle changes on net job creation cancel out—net job loss
in downturns is offset by net job gains in booms. Thus, in the long run net job creation is
determined by the characteristics of the labor market. Over time, a growing population leads to
rising employment, or net job creation. In 1950, the U.S. population equaled 151.3 million and
nonfarm employment equaled 45.3 million; in 2000, the U.S. population equaled 281.4 million
and employment equaled 131.8 million. (The fraction of the population employed in 2000 was so
much higher than in 1950 primarily because of the entrance of women into the labor force.)
What proportion of the population is employed—both the labor force participation rate and what
economists refer to as the “natural rate” of unemployment or NAIRU (non-accelerating inflation
rate of unemployment)—is determined in the long run by demography (e.g., younger workers
have higher unemployment rates), social norms (e.g., the large scale entrance of wives into the ²⁷
working force), and policy (e.g., welfare reform). In the long run, the rate of net job creation
will equal the rate needed to keep the economy at the natural rate of unemployment. Sometimes
the rate of employment growth will change because the natural rate is changing. For example,
from the 1970s until the early 1990s, the average unemployment rate was higher than in the

²⁷
^For^m^o^r^e^inf^o^r^m^a^{tion, s}^e^e^C^R^S^R^e^por^{t R}^L³²²⁷^4,^A^C^h^an^gin^g^N^a^tur^a^l^Rate^{of U}ⁿ^e^m^p^l^oy^m^e^nt:^P^o^li^c^y^I^s^s^u^e^s^{, by}
^Ma^r^c^L^a^bonte^.

1950s or 1960s for reasons that cannot be explained by the business cycle alone. Changes in
demography, social norms, and policy, as well as the coinciding productivity slowdown, all
played a role in the increase, but there is little consensus among economists on the relative
importance of each factor. Because changes in the natural rate are gradual, an increase in the
natural rate is more likely to be associated with a slower rate of net job creation than net job loss,
all else equal. At other times, the employment growth rate will change in order for the actual
unemployment rate to return to the NAIRU. Since unemployment rates in the most recent
expansion never fell to 1990s levels, part of the labor market sluggishness from 2001 to 2003
may have been caused by a return to the natural rate after unemployment was held below it in the
late 1990s.

Most economists agree that net job loss is an undesirable phenomenon, and recommend that
public policy be used to offset it. Policymakers can use expansionary monetary policy (lower
short term interest rates by the Federal Reserve) or expansionary fiscal policy (an increase in the ²⁸
budget deficit) to stimulate aggregate spending and offset net job loss. If used properly and
prudently, these policy tools can theoretically minimize net job loss. Unfortunately because of
policy lags in recognition, implementation, and effectiveness, fiscal and monetary policy will
probably never be conducted effectively enough to eliminate recessionary periods of net job loss.
Direct job creation programs have been used by the government in past recessions to stem net job
loss, but from an economic perspective, these policies have a similar effect to any expansionary ²⁹
fiscal policy, and they also are prone to implementation lags.
Policies that impede gross job loss (e.g., regulatory restrictions on dismissal or layoffs) may seem
to be a desirable way to limit net job loss at first blush. However, such policies could have the
unintended effect of making firms reluctant to take on new workers, because a firm would not be
able to subsequently reduce its workforce easily if the need for the new workers proved to be only
temporary. As a result, gross job gains could decline; if gross job gains declined by more than
gross job loss declined, net job creation would decline. This suggests that attempts to limit gross
job loss could be counterproductive. Because gross job flows are, on the whole, caused by the
reallocation of resources to their most efficient use, policies to impede gross job loss would also
likely have adverse consequences for growth and efficiency. Helping job losers make the
transition into a new employment situation is a less costly alternative, and one that is compatible
with an efficient, dynamic labor market.

²⁸
^For^m^o^r^e^inf^o^r^m^a^{tion, s}^e^e^C^R^S^R^e^por^{t R}^L³⁰³⁵^4,^M^o^ne^tar^y^P^o^lic^y^and^the^Fe^de^r^a^{l Re}^s^e^r^v^e^:^C^u^r^r^e^{nt P}^o^lic^y^a^nd
^Cond^itio^ns^,^by^Ma^r^c^La^bonte^;^aⁿ^{d C}^R^S^R^e^p^o^r^t^R^L³¹²³^5,^T^h^{e E}^c^on^o^mⁱ^c^{s o}^f^t^h^{e F}^e^d^e^ra^l^B^u^dg^et^Defⁱ^ci^t^{, by}^B^r^ia^{n W}^.
^Ca^she^ll.
²⁹^For^a^his^t^or^y^of^{job c}^r^e^a^tioⁿ^pr^og^r^a^m^s^{, s}^e^e^C^R^S^R^e^por^{t R}^L³¹¹³⁸^,^C^oun^te^r^c^y^c^lic^{al J}^{ob C}^r^e^a^ti^on^Pr^o^g^r^a^m^s^of^the
^Po^st-W^o^r^ld^W^a^r^{II Era}^{, by}^Lⁱ^nda^L^e^vⁱ^ne^.

International comparison confirms this view.³⁰ The Organization of Economic Cooperation and
Development (OECD) ranked countries on a scale of zero to six based on regulatory restrictions
on dismissal from regular employment, temporary employment, and mass layoffs. As seen in
Table 2, countries with little protection such as Switzerland, Japan, Australia, New Zealand, and ³¹
the United States had low unemployment rates, between 3.6% and 5.3% in 2007. Countries with
greater protection had a mixed experience: some small countries like Austria, the Netherlands,
and Norway kept unemployment low, but the four large countries with the most protection ³²
(Germany, France, Italy, and Spain) had relatively high unemployment rates. The
unemployment rate in the United States was lower at the trough of the recession than the lowest
level many of these countries attained at the peak of their business cycles.
Table 2. International Comparison of
Employment Protection Policies in 2003
^Regular^Temporary^Mass^Overall^{Unemployment Rate}
^Employment^Employment^Layoffs²⁰⁰⁷
^{Low Unempl}^{oyment Countri}^es
^{Norway 2.3}^2.9^2.9^2.6^2.6%
^{Netherlands 3.1}^1.2^3.0^2.3^3.2%
^{New Zealand}^1.7^1.3^0.4^1.3^3.6%
^Switzerland^1.2^1.1^3.9^1.6^3.6%
^{Denmark 1.5}^1.4^3.9^1.8^3.8%
^{Japan 2.4}^1.3^1.5^1.8^3.9%
^{Australia 1.5}^0.9^2.9^1.5^4.4%
^{Austria 2.4}^1.5^3.3^2.2^4.4%
^{United States}^0.2^0.3^2.9^0.7^4.6%
^United^{1.1 0.4}^2.9^1.1^5.3%
^Kingdom
^{Moderate Unemployment Countries}
^{Canada 1.3}^0.3^2.9^1.1^6.0%
^{Italy 1.8}^2.1^4.9^2.4^6.1%
^{Sweden 2.9}^1.6^4.5^2.6^6.1%

³⁰
^Se^e^a^l^s^o^B^a^r^t^H^o^bijn^aⁿ^d^A^y^se^g^u^{l Sa}^hi^{n, “}^J^o^b^Fi^ndi^ng^aⁿ^d^Se^pa^r^a^tⁱ^{on R}^a^te^sⁱⁿ^t^h^e^O^E^C^D^,”^Fe^de^r^a^l^R^e^s^e^r^v^e^B^a^nk^of
^N^e^w^Y^o^r^k^{, Sta}^f^f^R^e^por^{t N}^o^.²⁹^8,^A^u^g^u^s^t²⁰⁰^{7. T}^h^e^s^t^u^d^y^f^ound^th^a^t^Aⁿ^g^l^o-^Sa^x^{on c}^o^uⁿ^tr^ie^s^ha^v^e^sⁱ^m^ila^r^job-
^s^e^pa^r^a^{tion r}^a^te^s^to^ot^he^r^O^E^C^D^c^oun^tr^ie^s^,^b^u^t^ha^v^e^m^u^c^h^hig^h^e^r^jo^b-^fⁱ^nding^r^a^te^s^.^T^h^e^Uⁿ^ite^{d Sta}^t^e^s^ha^{d t}^h^e^hig^h^e^s^t
^job-fⁱ^nd^ing^ra^teⁱⁿ^the^sa^m^p^le^.
³¹^Uⁿ^e^m^ploy^m^e^{nt r}^a^te^s^{in a}ⁿ^y^gⁱ^{ven y}^e^a^r^m^a^y^be^inf^l^ue^nc^e^d^by^the^b^u^sⁱ^ne^s^s^c^y^c^l^e^.^A^{ll c}^ountrⁱ^e^sⁱⁿ^the^s^a^m^p^le^ha^d
^g^r^ow^ing^e^c^onom^ie^s^f^r^o^m^{2004 to 200}^{7, how}^e^v^e^r^.
³²^T^h^e^s^t^r^ong^e^s^{t c}^o^r^r^e^la^tion^s^e^e^m^s^{to be}^be^tw^e^eⁿ^une^m^p^loy^m^eⁿ^{t a}^nd^r^e^s^t^rⁱ^c^tions^oⁿ^te^m^por^a^r^y^em^plo^y^m^e^{nt pol}^ic^ie^s
⁽^w^{ith the}^e^x^c^e^p^tion^of^N^o^r^w^ay^,^w^hos^e^uⁿ^e^m^ploy^m^e^{nt r}^a^te^is^c^u^r^r^eⁿ^tly^be^ing^k^e^{pt low}^by^the^eⁿ^e^r^gy^boom⁾^.^Ma^ny
^obse^r^v^e^{rs ha}^v^e^c^r^e^d^ite^{d the}^Ne^the^r^la^{nds’ suc}^c^e^{ss w}^{ith lo}^w^une^m^p^loy^m^eⁿ^{t to its f}^e^w^re^stric^{tions o}ⁿ^te^m^pora^r^y
^em^ploy^m^e^nt.

^Regular^Temporary^Mass^Overall^{Unemployment Rate}
^Employment^Employment^Layoffs²⁰⁰⁷
^{High Unemployment Countr}^ies
^{Belgium 1.7}^2.6^4.1^2.5^7.5%
^{Portugal 4.3}^2.8^3.6^3.5^8.0%
^{France 2.5}^3.6^2.1^2.9^8.3%
^{Spain 2.6}^3.5^3.1^3.1^8.3%
^Germa^{ny 2.7}^1.8^3.8^2.5^8.4%
^Source:^OECD,^Empl^{oyment Out}^look^{, 20}^{04, Table 2.A2.4;}^OECD,^{Economic Outlook}^{, 2}⁰⁰^{8, Anne}^{x T}^{able 14.}
^Note:^{Measured on a scale of}^{0 t}^o^{6, with 0 e}^{qual to the least em}^{ployment p}^rotection.
If policies to forestall gross job loss are deemed to have too high an efficiency cost, what role can
policy play? Two perspectives can be used to answer that question. One perspective would view
the labor reallocation issue as a purely social problem. That is, allowing gross job loss to occur
with no impediment from the government may be an economically desirable outcome as long as
it is cancelled out by job creation, but the situation creates social problems that government may
wish to tackle on non-economic grounds. These social problems include poverty, psychological
problems, crime, alcoholism and substance abuse, the undermining of families and communities,
and so on that reportedly increase when gross job loss occurs. Economic analysis provides little
guidance on the best role for the government to play in tackling these non-economic problems.
Alternatively, an economic perspective would ask if any market failure is associated with gross
job loss, and what role the government can play in potentially rectifying that market failure.
Although it can be argued that gross job loss poses no market failure in and of itself, a persuasive
argument can be made that there are market failures that prevent individuals from efficiently ³³
insuring themselves in the private market against the risks posed by gross job loss. To a
considerable extent, the possibility of gross job loss is a risky event beyond a worker’s control,
such as adverse changes in the business cycle, tastes, technology, or trade patterns. One would
expect a worker to be willing to use some of his income to privately purchase insurance against
those risks, just as individuals insure against the risk of death, fire, health problems, and so on.
Yet the limited use of private unemployment insurance to supplement government-provided
insurance suggests that market failures may significantly hamper the functioning of the private
market.
All insurance markets are hampered by two important market failures—adverse selection and
moral hazard. A persuasive argument can be made that unemployment insurance may be more
adversely affected by both types of market failure than most other types of insurance. Adverse
selection is caused by asymmetric information: buyers of insurance know more about their
riskiness than sellers. As a result, only buyers with higher risks will tend to purchase insurance
because they are more certain that the benefit of the insurance will exceed the cost. This pushes
up the price of insurance and hampers insurers’ efforts to pool risk. Adverse selection hampers

³³
^For^a^m^o^r^e^in-^d^e^p^th^dis^c^us^sⁱ^on^o^f^the^f^o^llowⁱ^ng^a^r^g^u^m^e^{nt, s}^e^e^,^f^o^r^e^x^am^ple^,^R^obin^B^o^a^d^w^a^y^a^{nd D}^a^vⁱ^{d W}^ilda^s^in,
^Public^Se^c^t^or^Ec^o^nom^ic^s^(B^ostoⁿ^:^L^ittle^{, Brow}^{n, a}ⁿ^d^Co.^,¹⁹⁸⁴^),^p^p^.4^77-⁴⁸^7.

efficiency in the market for unemployment insurance because some causes of unemployment are
beyond the worker’s control, and some are not. Since insurance firms cannot identify which
workers have a greater chance of losing their jobs because of the factors within their control, they
cannot efficiently pool the risks that workers do not control. Moral hazard occurs when an
individual’s behavior becomes more risky once he is insured. Moral hazard also drives the cost of
insurance above its efficient level. Moral hazard can occur in the unemployment insurance market
in two ways: it can cause insured workers to engage in behavior that is more likely to lead to job
loss, and once job loss has occurred it can make an insured worker less willing to take a new job
(because the worker can subsist on the income provided by the insurance).
Government provision of unemployment insurance solves the adverse selection problem by
making participation universal. As long as all workers are participating, insurance can be priced at
its efficient level, even though benefit will not match cost for any given worker. Government
mitigates, but does not eliminate, the moral hazard problem by making the insurance temporary
(normally 26 weeks in most states) and imposing eligibility restrictions (e.g., not providing
insurance when the worker has quit or been fired). The government has also tended to extend the
duration of insurance during economic downturns, since events beyond the worker’s control are a
greater source of job loss then. The private sector could use the same methods as the government
to mitigate moral hazard, but it could not prevent adverse selection.
If one accepts that government provision of unemployment insurance is more efficient than
private provision, then the policy issue is whether or not insurance is adequate or excessive at
current levels. Are workers adequately protected against the risk of gross job loss at existing
benefit levels? Should benefit levels or duration be increased or coverage be expanded since gross
job loss seems to be following an upward trend? Would more generous insurance reduce the
social problems associated with gross job loss? The tradeoff here is between both benefit and cost
to the individual (more generous insurance would require higher premiums) and cost to the
economy because of the moral hazard problem: as the insurance becomes more generous,
disincentives to maintain employment or seek new employment among the unemployed increase.
Government’s role in insuring workers against the risks associated with gross job loss can also be ³⁴
viewed through a broader prism than the unemployment insurance program. Disability
insurance insures against the risk of job loss due to physical incapacity. Trade Adjustment
Assistance (TAA), which offers extended unemployment benefits and job training, reduces the
risk that workers adversely affected by trade will be unable to find re-employment. (Some
policymakers have suggested that TAA be extended to workers in the service industry, given the
growing concern with offshore outsourcing.) Government programs such as COBRA (named
after Title X of the Consolidated Omnibus Budget Reconciliation Act of 1985) reduce the loss of
health care associated with job loss. Some would argue that income redistribution, in general, is a
form of income insurance, whether it takes the form of progressive taxation, the Earned Income
Tax Credit, the food stamps program, the Supplemental Security Income program, and so on.
Kletzer and Litan have argued that the government should implement a “wage insurance”
program so that workers who lose their jobs and are forced to take lower paid employment are

³⁴
^T^h^e^the^o^r^e^tic^a^l^u^nde^r^pⁱⁿ^ning^s^of^this^vⁱ^e^w^a^r^e^la^{id out i}ⁿ^E^d^g^a^r^B^r^ow^ning^a^{nd J}^a^c^que^le^ne^B^r^ow^ning^,^Pu^blic^Fi^na^nc^e
^aⁿ^d^t^h^e^P^rⁱ^{ce S}^y^st^em^(Ne^w^York^:^Ma^c^m^illa^{n P}^u^blishⁱ^ng^,¹⁹^79),^p.¹⁸⁸^.

directly compensated by the government.³⁵ Along these lines, the Alternative Trade Adjustment
Assistance Program for Older Workers was introduced in 2002. An eligible worker (over 50 years
old, earning less than $50,000, and meeting other criteria) can receive half the difference between
the wages received from reemployment and the wages received at the displaced job for up to two
years and payments up to $10,000. This program applies only to workers affected by trade, ^th
although the economic rationale for such a program could apply to all workers. Bills in the 110
Congress, such as S. 1330 (Senator Schumer) and H.R. 2202 (Representative McDermott), would
create broader wage insurance programs.
At the community level, fiscal transfers (differences between outlays received and taxes paid) that
change with economic conditions and government programs such as the Empowerment
Zone/Enterprise Communities Program provide what could be characterized as “insurance” for
the community as a whole against the economic effects of significant job loss.

Drawing a distinction between net job loss and gross job loss can help to inform the policy
debate. Net job loss is a serious economic problem that fiscal and monetary policy can be used to
mitigate. Although it has social costs, gross job loss is part of the normal functioning of a market
economy, and has the beneficial role of reallocating resources to their most efficient use when
tastes, technology, or comparative advantage changes. Even in expansions, gross job loss is
sizeable, between 6.5 million and 8.5 million per quarter from 1992 to 2000, but it is more than
offset by gross job gains.
Trade, trade deficits, offshore outsourcing, overseas investment, and economic restructuring all ³⁶
cause gross job loss. But in normal economic conditions, none typically causes net job loss,
according to theory and evidence. To see why, consider that they all have the same effect on
employment as technological advances. For example, the advent of the automobile caused gross
job loss in the horse buggy industry, but was more than offset by gross job gains in the rest of the
economy.
As the buggy example suggests, policies that impede gross job loss can have high efficiency
costs. The difference in the unemployment experience of countries with high barriers to job loss,
such as the high unemployment countries of Western Europe, and countries with low barriers,
such as the United States, offers some evidence that barriers to gross job loss can lead to lower
gross job gains, making such barriers ultimately self-defeating. However, public policies to
protect workers against the risks posed by gross job loss can be justified on both social and
economic grounds. If crafted properly, they have been shown not to reduce gross job gains, and
they can arguably raise efficiency by addressing market failures. For example, public provision of
unemployment insurance helps overcome moral hazard and adverse selection problems in that

³⁵
^L^o^ri^Kl^et^{zer an}^d^Rob^e^rt^Lⁱ^t^aⁿ^,^“^A^P^r^escri^p^tⁱ^oⁿ^t^o^Relⁱ^e^{ve W}^o^rker^Aⁿ^xi^et^y^,^”^Institute^{for In}^te^rnati^o^nal^Ec^on^omic^s^,
^Polic^y^B^r^ief^01-^2,^Fe^br^ua^r^y^2001.
³⁶^Bure^a^u^of^L^a^{bor Sta}^tistic^s^da^taⁱ^ndic^a^t^e^tha^t^tra^d^e^aⁿ^{d o}^u^ts^o^u^rc^ing^c^a^u^se^re^la^tiv^e^l^y^little^g^r^{oss job l}^o^{ss, a}^t^le^a^s^{t in t}^h^e
^f^o^r^m^of^ex^te^nde^{d m}^a^s^s^lay^o^f^f^s^.

market. The challenge for policymakers going forward is to find the right balance between
mitigating risk and maintaining market dynamism in an increasingly fluid labor market.
^{Marc L}^a^bon^t^e
^Specialis^{t i}ⁿ^Macroecon^o^mⁱ^c^P^o^licy
^m^l^abon^t^e^@^c^rs^.l^oc.g^o^v^{, 7-}⁰⁶⁴⁰