You are on page 1of 11

In Debt as Sweat, Ishac Diwan argues that the relative mobility of capital has resulted in

an increased frequency and intensity of financial crises that periodically redistributes income
from labor to capital. Diwan finds three empirical regularities that support this conclusion: 1) A
tendency for labor share to fall during a financial crises, recovering only partially in non-crises
periods 2) A set of country specific characteristics that consistently determine the magnitude of
labor share decline during a crisis and 3) A secular fall of the labor share in countries that
experience numerous financial crises.
In a constructive manner, the paper layers conclusions to hone in on and test specific
mechanisms that are suspected of distributing wealth from labor to capital. The paper is divided
into three parts. First, the paper examines regional differences in labor share decline. From these
region specific results, the aforementioned empirical regularities are strongly supported. In the
second part, the paper attempts to find an explanation for these empirical regularities and
assesses four conceptual reasons: (i) Capital Mobility and Labors Race to the Bottom (ii)
Unequalizing Risk Sharing (iii) Public Sector Led Crises (iv) Private Sector Led Crises. To test
these conceptual ideas, a new regression attempts to explain labor shares relation to the overall
loss in an economy and to the economys underlying bargaining structure. A set of four variable
categories emerge: government policy, financial structure, change in GDP, and economic
characteristics. Each category has at least one determinant that reveal the specific mechanisms
through which labor bails out capital during a financial crises. The third and final part of the
paper takes into account hysteresis. The test of hysteresis tests the dependency of a current
output on past and current inputs. The conclusion reached is that countries whos labor share is
impacted by previous financial crises have labor shares that are particularly sensitive to financial
crises. Change is accelerating over time, and a potentially ominous point is subtly raised: as the

1 of 11

forces of change amplify the strength of faceless capital vis-a-vis the labor masses, does the
human cost of the subsequent increased vulnerability of loss outweigh the accrued gains of
capital mobility?
The following is a detailed summary of Diwans argument. For the purposes of cogency,
the summary will underscore the mechanisms through which globalization engineers income
redistribution from labor to capital. The following will establish why labor as a share of GDP is
an acceptable variable to use to measure inequality. Second, the summary will define what a
financial crises is and why using it as an event within the regression is important to highlight the
effects of different explanatory variables. Third, the summary will explain how a regional break
down of Labor Share in response to a crisis produced the aforementioned empirical regularities.
Fourth, the summary will use these empirical regularities as a springboard to explore conceptual
stories that could possibly provide an explanation behind the downward trend. This develops a
testable hypothesis. Fifth and finally, different determinants will be used in a regression to
understand how capital is redistributed.
Diwan opens with the question of whether globalization hurts the common man.
Conventional wisdom hypothesizes that the relative mobility of capital vis-a-vis labor has caused
two circumstances to arise: (i) labor will bear the brunt of the burden of negative shocks because
the threat of capital flight enables investors to demand the international market rate plus a risk
premium and (ii) capital mobility forces labor to compete on a global marketplace depressing
wages; this phenomenon is called race to the bottom (page number. Diwans findings are a
challenge to the conventional globalization thesis which elegantly constructs the liberalized order
as a slow monotonous machine that increases inequality between the rich and poor. The
innovative use of labor share as a variable of interest allows for the observance of short term

2 of 11

distributional shifts in income. The response variable is the percentage of the countrys income
(GDP) that goes to labor, specifically non-resident and resident household income, this is known
as labor share. This analysis establishes that the battlefield of globalization is characterized by
short sporadic distributional fights between capital and labor. When evaluating the effects of
globalization and attempting to correct its excess, it behooves one to acknowledge how
mechanisms enhance inequality.
To examine the relationship between labor and capital, Labor share of GDP is the
proximate variable. There are several problems with using other traditional measures of
exploitation such as poverty rates, income inequality, or return to education. Non-proximate
variables do not adequately examine the relationship between capital and income: the
relationship is easily obfuscated by various confounding variables, survey data is limited because
it cannot measure short term shifts in data, and measures of consumption data are obscured by
consumption smoothing. Consumption smoothing occurs when the government responds with
taxes, wealth transfers, or safety nets to a shock. Thus, labor share of GDP is the best way to
measure short term distributional shifts between labor and capital.
There is an important limitation worth noting about Labor share of GDP. The variable is
estimated on the basis of enterprise and governmental accounts, but is limited because it does not
take into account the informal or self-employed economic sectors, leading to a potential
overestimation of a financial crises impact on true labor share. This is a major limitation to the
studys approach as crises result in a migration to these sectors. It is of significant importance
because most developing countries have large informal sectors. Generally, these countries are the
most sensitive to the type of financially induced crises that are examined in this paper. The paper

3 of 11

attempts to correct for this weakness by controlling for the size of the countrys rural sector in its
regression.
The papers supposition is that financial crises provide an opportunity to examine how
distribution occurs. The study defines a financial crises event as a 25% decrease in the nominal
exchange rate. Downward pressure on the nominal exchange rate indicates that capital is trying
to withdraw. The capital in question, short term financial claims on the state or short term
deposits in local financial institutions, is sensitive to the international market rate. The
international market rate is important because the state and banks have to compete against it. If
investors suspect a rate change in response to a negative shock, divestment from the respective
currency can occur. This can result in a currency run that spirals into a financial crises.
The study separates crises and non-crises regimes, this eliminates the concern of
misleading average behaviors (3). Simply put, using financial crises as an event within the
regression will allow us to see how labor share is impacted inside and outside of a crisis and
ultimately test whether financial crises are in fact distributional fights over wealth. In addition,
utilizing time trend and hysteresis analysis, the regression will reveal how labor shares reaction
to a financial crises is impacted by the previous number of crises. That result can then give us an
idea about how globalization increases inequality.
The first regression in the study is labor share on a time trend and several crises dummy
variables. The second regression controls for regional variation. The regions examined are Latin
America, Africa, Middle East, Asia and the OECD. The regression results are reported in Table 1.
In the first regression, it is observed that labor share exclusively falls within a financial crises
event and recovers when . In the second regression, controlling for different regions allows us to

4 of 11

see that some regions are more susceptible to a decline in labor share than others. To understand
the implications of these findings, further exploration of the data is warranted.
Percentage declines in labor share are relative to GDP held constant. Table 2 reports that
World Labor share falls 6.1 percentage points of GDP per crises. However, that number varies
regionally: Asia is 2.3, Middle East is 4.2, Africa is 5.92, and Latin America is 6.72. Again, it is
important to understand that these numbers assume that GDP is held constant in a financial
crises. If there is a decline in GDP in conjunction with a financial crises that can amplify the loss
to labor: the overall pie gets smaller and the financial crises distributes wealth away from labor.
That results in bigger total losses.
By region, Table 2 calculates the net loss, transitory loss, and permanent loss resulting
from a crisis. Net loss reflects the reality that in non-crisis years labor share increases. Permanent
losses are accumulated losses added up till the end of a crisis. Transitory losses are also
accumulated losses over the crisis period that are non-permanent.

When permanent and

transitory losses are added together, they reveal the total wealth transfer from labor to capital
during a crisis. To provide context to these numbers and provide a sense of scale, let us take the
Latin American region as an example. On average, Latin Americas net loss is 7.4 points of GDP
per crises. While total accumulated labor loss during a crisis is 36.8 percent. This is a massive
amount. The studys initial regression demonstrates that financial crises disproportionally hurt
the working class, even if labor share eventually recovers.
From the regression, two empirical regularities emerge. First, labor share tends to
massively decrease during a financial crises. The majority of the loss is transitory, or nonpermanent, but recovery is only partial. Generally, labor share will experience a net loss from a
financial crises. Our examination of the Latin American region demonstrates this conclusion.

5 of 11

Second, over time, labor share decreases. However, that decrease seems to only occur during a
financial crises. These are meaningful peculiarities because the findings raise further questions
about what mechanisms financial crises elicit to transfer income from labor to capital, and also,
what is causing downward pressure on labor share over time.
To understand these empirical regularities, Diwan attempts to provide conceptual links
to establish financial crises relationship with labor share. He first discounts the neoclassical
model of growth as an inadequate framework for analysis. He argues that it does not allow for
fluctuations in labor share. Additional readings suggest that this is because the neoclassical
growth model features a Cobb-Douglas production function which assumes that the labor share
is constant. Diwans paper is an empirical refute to this notion. Thus, the paper looks to 4
conceptual stories that can explain the wealth transfer. Utilizing local and national political
structures, the stories are used to develop a testable hypothesis. Those four stories are labeled as
follows: (i) Capital Mobility and Labors Race to the Bottom (ii) Unequalizing Risk Sharing (iii)
Public Sector Led Crises and (iv) Private Sector Led Crises. The following is a brief explanation
of each of these conceptual links. The ultimate hypothesis that stems from this discussion is that
labor share should be related to the overall loss in the economy and to the countrys underlying
bargaining structure.
Capital mobility and labors race to the bottom states that when capital can move easily,
labor share will fall in rich countries and rise in poor countries because labor is cheaper in
developing countries. However, acting as a countervailing effect, local capital will invest in the
most efficient labor markets, this sometimes does not exist in developing countries. Capitals
mobility will enhance the interplay of these two dynamics and depress labors wages. Diwan
points to various defunct economic labor systems of the 20th century such as communism, state

6 of 11

led growth, and aid led growth as examples of less efficient systems that were corrected by
capitals enhanced mobility. The abandonment of these systems required a reallocation of
resources from labor to capital.
Unequalizing risk sharing is the idea that developing countries lack mature financial
institutions which results in an inability to correctly allocate risk. The study had already
established that capital mobility puts more pressure on other factors of production to bear more
risk. When inevitable macroeconomic shocks occur, such as oil shocks, food prices fluctuations,
or other events, developed nations have the institutional strength to ensure smoothing of
consumption. Institutions such as mature equity markets, flexible labor markets, sound debt
management practices, and progressive tax policies enable a quick rebound because they
socialize risk. Developing economies rely on state funds to bail out the troubled sector. The
result is an increase in public debt to cover up short term macro economic losses. This causes
phenomenons such as debt overhang to occur. Debt overhang happens when existing debt
obligations make it onerous to borrow additional money, even when that money is guaranteed to
pay itself off. This sets the state up for inevitable and potentially harsher crises down the line.
Those crises end up requiring painful corrections that often are placed at the feet of the nations
citizens.
These two explanations suggest capital mobility has increased the number and intensity
of financial crises, it also provides an explanation for why labor share is decreasing. However,
they do not provide a conceptual link that explains the mechanisms behind why financial crises
have a disproportionate impact on labor. Diwan acknowledges this and proceeds to categorize
financial crises by origin; either classifying them by a Public Sector Led Crises or a Private
Sector Led Crises. The origin based approach can elucidate the different mechanisms that are

7 of 11

used to resolve the crises and perhaps can reveal the bailout mechanism by which labor transfers
wealth.
In a Public Sector Led Crises, Diwan says that a public sector solvency crises triggered
by internal or external shocks can lead to a financial crises. A common variable across financial
crises is the drying up of funds a government has access to. Investors pull money out of the
domestic currency because of a fear of inflation. Investors view inflation as a tax, or a way for
the government to water down its debt obligations. Capitals increased mobility has blunted this
instrument of choice, and so forces the government to either raise taxes, cut spending, or default
on debt. During the 1980s, the primary way many Latin American countries resolved their debt
crises was by enacting public-sector wage reductions. The public sector of a typical Latin
American country comprised of about 10 percent of overall GDP, the magnitude of a typical
wage
reduction enabled the government to pay off its debt. This is a specific mechanism that shows
how labor bails out capital. It may belabor the obvious, but a reduction in previously agreed upon
public sector wages to meet creditor obligations is a transfer from labor to capital. In addition,
there is interplay between the private sector and the public sector, as public sector wages decline,
that depresses wages in the private sector because there are more workers available. This is
referred to as a contagion, whereby the harmful effects in one sector spillover to another.
A private sector led crises can arise when markets take on excessive risk because of
expectations of future bailouts. In the past, governments have shown a willingness to maintain
financial stability by reducing risk and guaranteeing private loans. The mechanism of
privatizing gains and socializing losses transfers wealth from ordinary citizens to corporations

8 of 11

and banks. That in turn leads to the aforementioned debt overhang problem, further exacerbating
the wealth transfer.
Finally, the private sector through corporate clubs coordinates wage reductions with
unions to restore profitability and by extension, solvency within the banking system. During a
crisis, unions do not have much bargaining power. They would rather have their members be
employed with a smaller check than unemployed. This is yet another mechanism by which labor
can bail out capital. Corporate profitability enhances bank sheet balances and can help resume
normal banking activity.
These conceptual links demonstrate that a countrys underlying bargaining structure
between labor and capital and the extent of a crisis have the potential to be strong determinants
of labor share. The paper runs a regression utilizing the following determinants - (i) economic
policies: governmental deficit and capital controls (ii) financial structure: other domestic liquid
claims, illiquid claims, equity (iii) change in GDP and (iv) economic environment: time trend,
size of trade, size of the rural sector.
Rather than explore all the findings of each regression (many are performed to establish the
importance of including several variables), this summary will highlight specific points that
support the core article findings. The most interesting characteristic about this regression is that
it explores how the determinants act in two different regimes: crisis and non-crisis. It is worth
noting that these determinants are largely considered on a short-term basis. Meaning, these
determinants are not tracked over time. The paper does entertain medium term considerations by
citing other papers, but this acts as complementary information and will not be included in this
summary.

9 of 11

Economic Policies are shown to have a strong impact on labor share. Capital controls are
especially effective at retaining labor share during a crisis. Predictably, government spending has
a positive relationship with labor share, for every point increase in government spending, labor
share increases by .15 points. This does not vary across crises regimes. However, fiscal deficits
does vary across regimes, in a non-crisis regime, a fiscal deficit is associated with an increase in
labor share. However, during a crisis, a fiscal deficit will hurt labor. This reflects the debt
overhang story stated earlier. Crises are either caused by debt or create new debt issues, either
way, a debt resolution often falls on the burden of ordinary citizens.
Financial structure is the main determinant of the "burden sharing between capital and
labor during a crisis". When external debt and M1 (measure of most liquid components of the
money supply) are high, that reflects enhanced capital mobility. If capital is mobile, the
bargaining structure is in favor of capital. This is confirmed by regression results. During a crisis,
high levels of mobile capital results in a decrease in labor share. If M2 and and M3 are larger
within an economy, that is associated with a positive relationship for labor share.
In a set of slightly surprising results, in a crisis, log(GDP), or GDP movement, is
negatively related to movement in labor share. This is telling. In a crisis, if GDP decline is larger
than labor share, the decline is not as big, if GDP decline is small, then labor decline is more.
This could corroborate the idea that if labor does not bail out the economy, then GDP decline is
protracted. Outside of a crisis, GDP fluctuations are positively related to labor share.
The last determinant that the regression tests for is economic environment. This is broken
down to population size, rural share and the extent of trade. Rural share is controlled in an
attempt to address the previously mentioned weakness of using labor share as a data source.
When economic decline occurs, populations tend to rely on agriculture and the informal sector

10 of 11

for income. Population size is found to be positively correlated with labor share, this suggests
that population size acts as a buffer against the negative effects of capital mobility. During a
crisis, trade is found to have a countervailing effect on labor share. However, this variable only
measures exports relative to GDP at the time of the crisis and is not reverse casual. Thus, it is not
a forgone conclusion that if trade were to increase that would necessarily result in an increase in
labor share.
The final analysis conducted is an analysis of the time trend. Through time, there is an
overall downward trend in labor share. Utilizing hysteresis in the regression provides an
explanation. Hysteresis controls for the number of past crises/years experienced by a country.
Once it was controlled for, the time trend reported a statistically significant but small negative
trend. The hysteresis coefficient was negative and was statistically and materially significant. The
number of past crises experienced by a country and Labor Shares resiliency is inversely
correlated. Diwan speculates that this is a result of a wear down effect. Labor loses ground that
it cannot make up, the power lost creates a cascading effect where it will be even more
susceptible during the next financial crisis.

<Did not include a conclusion for the summary portion of this paper, that will come with the
extension, which has yet to be written.>

Core Source:
Diwan, Ishac. "Debt as Sweat: Labor, Financial Crises, and the Globalization of
Capital."Debt as Sweat: Labor, Financial Crises, and the Globalization of Capital(n.d.):
n. pag. World Bank, 2001. Web. 27 Feb. 2015.

11 of 11

You might also like