Wages in a General Equilibrium Time Series Model of the US
Henry Thompson
Auburn University
April 2008
Theoretical wage effects of changing product prices and inputs have been examined under various
assumptions in general equilibrium models of production. The present paper is the first to estimate
wage adjustments in the context of this theory. The data cover the real wage, labor force, fixed
capital assets, energy input, and prices of manufactures and services in the US from 1949 to 2006.
Labor is revealed in the middle of the factor intensity ranking. The rising price of services has raised
the wage, as has the falling price of manufactures. Energy input has had a larger wage impact than
investment.
Contact information: Economics, 202 Comer Hall, Auburn University AL 36849, 334-844-2910, fax
5999, thomph1@auburn.edu
1
Wages in a General Equilibrium Time Series Model of the US
Wages endogenously adjust as markets clear in response to changing product prices and
factors endowments in general equilibrium models of production. Theoretical wage effects have
been analyzed under various assumptions but actual adjustments have not been examined in the
context of the general equilibrium model. The present paper estimates wage adjustment to
exogenous changes in prices of manufactures and services, and inputs of labor, capital, and energy in
the US from 1949 to 2006. The estimated difference equation is derived from the general
equilibrium model with three factors and two goods, and estimated coefficients relate to factor
intensity and substitution.
The general equilibrium model based on full employment and competitive pricing is a logical
extension of Walras (1874), Heckscher (1919), and Ohlin (1924) as formalized by Samuelson
(1953), Jones (1956), Chipman (1966), Jones and Scheinkman (1976), Takayama (1982), and Ethier
(1984) and reviewed by Jones and Neary (1984). The present paper applies the model with three
factors and two goods of Ruffin (1981), Suzuki (1983), and Thompson (1985, 1993).
The model provides the foundation for applied general equilibrium models based on
observed factor intensity and estimated substitution. Thompson (2005) reviews some simple direct
specifications. The implied wage relationship, however, has not been estimated. The present
variables are cointegrated and error correction models separate adjustment relative to the dynamic
equilibrium.
Estimated coefficients relate directly to parameters of the comparative static general
equilibrium model, including factor shares, industry share, and substitution elasticities. Coefficients
also provide a gauge of fundamental exogenous forces on the labor market. Practical issues
2
addressed by the present results include the wage effects of the falling price of manufactures due to
import competition and the rising price of increasingly exported services. Regarding issues of
changing factor endowments, the wage effects of immigration, altered investment, and decreased
energy input can also be analyzed.
1. An applied 3x2 general equilibrium model
Inputs of capital K, labor L, and energy E produce outputs of manufactures xM and services
xS in the present 3x2 general equilibrium model of production and trade. Full employment of labor
is stated L = ΣjaLjxj where L is the labor force and aLj the cost minimizing input of labor per unit of
good j. Differentiate and introduce factor cost shares θLj and substitution elasticities σik in the first
equation of the general equilibrium comparative static system (1) with exogenous changes on the
right hand side. The second and third equations in (1) reflect adjustment to exogenous changes in
capital K and energy E inputs.
Competitive pricing of product j is written pj = aLjw + aKjr + aEje where factor prices are the
wage w, capital rent r, and energy price e. Differentiate and use the cost minimizing envelope
theorem to derive the last two equations in (1) where industry shares λij are portions of factor i
employed in sector j.
The comparative static model stated in differences of natural logs is
σLL
σLK
σLE
θLM
θLS
dlnw
dlnL
σKL
σKK
σKE
θKM
θKS
dlnr
dlnK
σEL
σEK
σEE
θEM
θES
dlne
λLM
λKM
λEM
0
0
dlnxM
dlnpM
λLS
λKS
λES
0
0
dlnxS
dlnpS
=
dlnE
(1)
.
The system matrix is the Hessian of a constrained neoclassical national income maximization and
Chang (1979) shows its determinant ∆ is negative due to concavity. Cross price substitution
3
elasticities are symmetric σij = σji and homogeneity implies substitution elasticities sum to zero Σiσji
= 0.
The focus is wage adjustments to exogenous changes in prices and factor endowments. Let
one of these exogenous changes be nonzero and divide both sides of (1) by that change. Solve for
partial derivatives of the wage with Cramer’s rule to find
εwL ≡ δlnw/δlnL = θKEλKE/∆
εwK ≡ δlnw/δlnK = -θLEλKE/∆
εwE ≡ δlnw/δlnE = -θLKλKE/∆
(2)
εwM ≡ δlnw/δlnpM = -(λKSφ1 + λESφ2)/∆
εwS ≡ δlnw/δlnpS = (λEMφ2 – λKMφ1)/∆
where θEK ≡ θEMθKS – θESθKM, θEL ≡ θEMθLS – θESθLM, θLK ≡ θLMθKS – θLSθKM, λEK ≡ λEMλKS – λESλKM,
φ1 ≡ (θEK + θLK)σLE + (θLK – θEL)σKE, and φ2 ≡ (θLK – θEL)σKE – (θEK + θEL)σLK given homogeneity.
Factor intensities alone determine the signs of θKE, θLE, θLK, λKE, and by implication the first three
terms in (2). If energy is intensive in manufacturing relative to labo, for instance, θEL and λEL are
positive.
The own wage effect εwL is negative reflecting neoclassical concavity since θKE and λKE have
the same sign and ∆ < 0. Signs of εwK and εwE depend only on factor intensity but one must be
positive, an increase in either capital or energy raising the wage. A negative effect might not be
expected for either but output adjustments and substitution may not favor the wage. Signs of εwM
and εwS depend on factor substitution as well, and the sizes of all wage effects depend on the factor
substitution as well as intensity.
Partial derivative wage effects can be summarized in a single equation as
dlnw = (λKE(θKEdlnL – θLEdlnK – θLKdlnE) – φMdlnpM + φSdlnpS)/∆
(3)
where φM ≡ λKSφ1 + λESφ2 and φS ≡ λEMφ2 – λKMφ1. Empirical specification of the wage difference
equation in (3) is
4
dlnw = α1dlnL + α2dlnK + α3dlnE + α4dlnpM + α5dlnpS + ε
(4)
where ε is a white noise residual and time subscripts are omitted. Estimated coefficients such as α1 =
εwL = λEKθEK/∆ reveal effective factor intensity and perhaps information on substitution. Theoretical
expectations are a negative α1 and at least one positive sign for α2 and α3.
2. Difference stationarity
The data in Figure 1 are from the US National Economic Accounts of the Bureau of
Economic Analysis (2007) except Btu energy input E from the US Department of Energy (2007).
The wage w is derived from employee compensation averaged across the labor force L and deflated
by the consumer price index. The capital stock K is the deflated net stock of fixed capital assets.
* Figure 1 *
Price of manufacturing pM and services pS are price indices relative to the consumer price
index. Part of the 68% decrease in the price of manufactures is due to import competition. Quite the
opposite, the price of services increased 59% over the period. The relative price of services ps/pm
increased about five times and the output of services relative to manufactures almost 50% according
to output indices in the same data.
Difference stationarity is critical to estimating the difference equation (4). The log of the
price of services lnpS in Figure 2 is difference stationary by the Dickey-Fuller DFc test as reported in
Table 1. Energy input lnE is difference stationary by the augmented Dickey-Fuller ADF test adding
α2t and α3∆x-1. The wage lnw and capital input lnK are difference stationary in ADF(2) tests with
the added lag α4∆x-2.
* Figure 2 * Table 1 *
The labor force lnL has autocorrelated residuals and significant coefficients in all DF tests,
and the price of manufactures lnpM has positive own coefficients. These two series in Figure 3 are,
however, difference stationary with a 1975 structural break by the Perron (1989) test in Table 1.
* Figure 3 *
5
In the difference stationary tests there is no evidence of autocorrelation relative to the critical
Durbin-Watson statistic DW = 1.43 and no heteroskedasticity in residuals according to ARCH(1)
tests. The series are all I(1) processes with unit roots suggesting a reliable difference equation
regression in (4).
3. Wage difference and error correction models
The residual of the regression in levels can be utilized in an error correction model ECM if
the series are cointegrated and Table 2 reports the spurious regression. The series are cointegrated
according to the Engle-Granger EG (1987) test reported in the last column implying interdependent
time paths and suggesting an ECM.
* Table 2 *
The first row in Table 3 reports the estimate of the structural equation (4) as the difference
equation model
∆lnw = α1∆lnL + α2∆lnK + α3∆lnpM + α4∆lnpS + α5D + α6D∆lnL + α6D∆lnpM + ε
(5)
where D represents the 1975 structural break. Coefficients of D and its interactions are insignificant.
Coefficients are estimated comparative static arc elasticities with other exogenous changes held
equal to zero.
* Table 3 *
The own labor ∆lnL elasticity is negative and elastic. Both the capital ∆lnK and energy ∆lnE
elasticities are positive. The range of critical DW values for no autocorrelation is 1.43 to 2.57 and
the difference model falls in the indecisive range, casting some doubt on parameter significance.
Heteroskedasticity is revealed by the ARCH(1) test.
The theoretical model has no constant in (4) but adding a constant gauges other potential
influences on the wage. The constant also detrends the wage although that is not necessary for the
difference stationary series. The second row in Table 3 reports the difference model with a
6
significant constant and coefficients very similar to the difference model with no constant except a
much weaker labor force effect after the break.
The third row in Table 3 reports the ECM and its significant cointegrating effect.
Coefficients have values similar to the difference model. Coefficients of the structural break D and
its interactions are insignificant. Adjustment to the dynamic equilibrium reflected by the error
correction coefficient strengthens the net wage elasticity. Note, however, no explanatory power is
added relative to the difference model. Further, ambiguity in autocorrelation remains in the DW
statistic and ARCH(1) heteroskedasticity is stronger.
The ECM with a constant is reported in the last row of Table 3. Coefficients are similar to
the ECM with no constant although there is a much weaker labor force effect after the break.
Statistical properties are reliable and this ECM is included in the model comparison in Table 6.
The dynamic adjustment process is extended to lags of differences in the lagged error
correction model ECM-1 in Table 4. Relative to the ECM explanatory power increases and both
autocorrelation and heteroskedasticity disappear. Elasticities of labor, energy, and the price of
services are stronger, and energy input has a lagged effect. The capital elasticity is insignificant for
the first lag and weaker for the second lag relative to the ECM. The price of manufactures has no
wage effect. The structural break is significant interacted with ∆lnL implying a different labor force
effect after 1975. The cointegrating coefficient is over twice as strong in its second lag compared to
the ECM. Regressions with further lags reveal no significant effects, suggesting all wage adjustment
is complete after two years.
* Table 4 *
Each exogenous effect is separated into a difference and an error correction effect. For
instance, ∆lnL has a wage effect of -1.83 and an error correction effect of -0.53 x 0.69 = -0.37, the
two summing to -2.20. The ECM-1 model is included in the model comparison in Table 6.
7
The possible autocorrelation in the difference model suggests it can be utilized in an error
correction model in double differences, and the differences are cointegrated as indicated by that
Engle-Granger EG test. The Engle-Granger statistic is significant at-5.37* and there is no
autocorrelation with DW = 1.79 or heterskedasticity with ARCH(1) = 0.98. With no a constant term,
the coefficient of each double difference variable equals the difference coefficient in (4).
The first row in Table 5 reports the double difference model. There is no autocorrelation or
heterskedasticity, an advantage over the difference model. The labor endowment effect is weaker
after the 1975 break. The effects of capital and the price of manufactures disappear. A constant in
the second row of Table 5 does not affect results.
* Table 5 *
The third row of Table 5 reports the double difference error correction model ECM2 with the
residual of the difference model as the error correction variable. There is no autocorrelation or
heterskedasticity and explanatory power is reasonable. Elasticities of the wage, energy input, and
price of services are similar to other models but effects of the capital and price of manufactures
effects disappear. The break in labor input implies a weaker effect after 1975. The model with a
constant in the second row produces nearly identical coefficients.
The error correction effect is sizeable and effects of all differences feed through the error
correction process. As an example, the error correction effect of energy input is -0.72 x -0.74 = 0.53
making its total effect 0.65 + 0.53 = 1.18. Coefficients of this double difference ECM2 model are
included in Table 6.
4. Comparison of wage elasticities
Labor is revealed as the middle factor in the intensity ranking by the input elasticities.
Estimated signs (- + +) of wage elasticities (εwL εwK εwE) imply the terms (θKE θLE θLK) in (2) have
signs (– – +). There are two factor intensity rankings consistent with these signs are
θEM/θES > θLM/θLS > θKM/θKS
(6)
8
and its reverse. The high input of energy in manufacturing suggests (6). The larger size of εwE
relative to εwK implies θLK > θEL which means θLM/θLS is closer to θKM/θKS than θEM/θES. In other
words, energy is revealed as very intensive in manufacturing.
One of the cross price elasticities (σKE, σLE, σLK) may be negative indicating technical
complements but otherwise factors are substitutes. There is a literature on whether energy and
capital are complements as they appear in some time series analysis and the present wage elasticities
εwM < 0 and εwS > 0 are consistent with σKE < 0 in (2).
Wage elasticities are summarized in Table 6 for the error correction model ECM in Table 3,
the ECM with a lag in Table 4, and the double differenced ECM2 in Table 5. Models without
constants are closer to theory and utilized in Table 6 and averages with constants are similar.
Coefficients are generally similar and not far out of line with the spurious model in levels, point
estimates of the same elasticities. For discussion, consider their average in the last row of Table 6.
* Table 6 *
Sizes of wage effects can be gauged by means and standard deviations of changes. For
instance, the mean change 1.4% of the labor force ∆lnL and its standard deviation 0.5% suggest the
range of changes (1.9%, 0.9%). The average labor elasticity -2.28 before the 1975 break implies an
average wage adjustment of -3.2% and the range (-4.3%, -2.1%). After 1975 the average wage
elasticity of -0.66 implies an average wage reduction of 0.9% and the range (-1.3%, -0.6%). Steady
increases in the labor force put continuous downward pressure on the wage, reflecting a competitive
labor market.
Immigration contributes to the increase in the labor endowment although the share is difficult
to pin down due to a lack of data. Assuming half of the immigrants obtaining legal status and
located aliens enter the labor force, immigration accounts for up to 1/2 of the yearly increase in the
labor force. At the mean, that assumption would imply annual immigration lowered the wage ceteris
paribus by -1.6% before 1975 and -0.5% since.
9
The mean capital change ∆lnK of 3.5% and its standard deviation 2.2% imply an average
wage increase of 1.0% and the range (1.6%, 0.4%) with the average 0.28 elasticity. The mean
energy input change of ∆lnE 2.0% and its standard deviation 2.8% imply an average wage increase
of 2.2% and the range (5.3%, -0.9%) with the 1.10 energy elasticity. Increases in the labor
endowment are more than offset after 1975 by increased labor demand due to investment and
increased energy input. Before 1975, however, competition in the labor market dominated on
average depressing the wage holding output prices constant.
The mean change in the price of manufactures ∆lnpm -2.0% and standard deviation 1.7%
imply the range (-0.3%, -3.7%). The elasticity of -0.36 implies an average wage adjustment of 0.7%
and range (1.3%, 0.1%). While the positive wage effect of a falling price of manufactures may seem
paradoxical, output of labor intensive services was increasing.
The mean and standard deviation of the change in the price of services ∆lnps are both 0.8%
implying the range (1.6%, 0%) and its 1.31 elasticity implies an average wage effect of 1.0% and
range (2.1%, 0%). These annual manufacturing and services price effects are consistent with energy
intensive manufactures and labor intensive services comparing these two inputs as in the intensity
ranking (6).
5. Conclusion
Empirical results reveal labor as the middle factor relative to capital and energy, and energy
as very intensive in manufacturing. The ceteris paribus wage effect of the typical annual increase in
the labor force is about -4% before 1975 but reduces to about -1% since. Immigration accounts for
perhaps half that decrease in the wage.
The rising price of services raises the wage about 1% annually and the falling price of
manufactures raises the wage almost that amount. The underlying consistent expansion of the
service sector has supported the wage as the relative price of services increased over five times.
10
The wage increases by an average of 1% annually due to investment and over twice that
amount due to increased energy input. The expected continuing rising price of energy in the future
does not bode well for labor. Subsequent models and empirical analysis can look closer at the
relationship between the wage and energy input.
Regression results are consistent with substitution between all inputs or with complementary
capital and energy. Subsequent models or data may reveal more about effective substitution.
Estimated substitution parameters vary across regions and time, and the present approach may
contribute to the applied production literature.
The present sort of empirical analysis has potential to refine general equilibrium models of
production. The model can include other factors of production including labor skills, natural
resources, and sector specific capital or other inputs. Outputs can be disaggregated. The various
equations of the general equilibrium production model, including adjustments in all factor prices and
outputs, can be estimated. Implications of the various theoretical assumptions in the literature can be
compared. Adjustment processes can be compared across countries and time periods.
11
References
Bureau of Economic Analysis (2007) National Economic Accounts, Department of Commerce
webpage, www.bea.gov/national/nipaweb/SelectTable.asp?Selected=N
Chang, Winston (1979) “Some Theorems of Trade and General Equilibrium with Many Goods and
Factors,” Econometrica 47, 709-26.
Chipman, John (1966) “A Survey of the Theory of International Trade: Part 3, The Modern Theory,”
Econometrica 34, 18-76.
Department of Energy (2007) Energy Overview, Energy Information Agency webpage,
www.eia.doe.gov/emeu/aer/overview.html
Dickey, David and Wayne Fuller (1979) “Distribution of the Estimates for Autoregressive Time
Series with a Unit Root,” Journal of the American Statistical Association 74, 427-31.
Engel, Robert and Clive Granger (1987) “Cointegration and Error-Correction: Representation,
Estimation, and Testing,” Econometrica 55, 251-76.
Ethier, Wilfred (1984) “Higher Dimensional Issues in Trade Theory,” in Ron Jones and Peter Kenen (eds)
Handbook of International Economics, vol. 1, Amsterdam: North Holland, 131-84
Heckscher, Eli (1919) “The Effect of Foreign Trade on the Distribution of Income,” Ekonomisk
Tidskrift.
Jones, Ron (1965) “The Structure of Simple General Equilibrium Models,” Journal of Political
Economy 73, 57-72.
Jones, Ron and Neary, Peter (1984) “The Positive Theory of International Trade,” in Handbook of
International Trade, vol. 1, Ron Jones and Peter Kenen (eds.), Amsterdam: North Holland.
Jones, Ron and José Scheinkman (1977) “The Relevance of the Two-Sector Production Model in
Trade Theory,” Journal of Political Economy 85, 909-35.
Ohlin, Bertil (1924) The Theory of Trade, translated in Harry Flam and June Flanders, HeckscherOhlin Trade Theory, MIT Press, 1991, 73-214.
Perron, Pierre (1989) “The Great Crash, the Oil Shock, and the Unit Root Hypothesis,”
Econometrica 57, 1361-1401.
Rassekh, Fahad (1992) “The Role of International Trade in the Convergence of Per Capita GDP in
the OECD, 1950-1985,” International Economic Journal 6, 1-15.
Ruffin Roy (1981) "Trade and Factor Movements with Three Factors and Two Goods," Economics
Letters 7, 177-82.
12
Samuelson, Paul (1953) “Prices of Factors and Products in General Equilibrium,” Review of
Economic Studies 21, 1-20.
Stolper, Wolfgang and Paul Samuelson (1941) “Protection and Real Wages,” Review of Economic
Studies 9, 58-73.
Suzuki Katsuhiko (1983) “A Synthesis of the Heckscher-Ohlin and Neoclassical Models of
International Trade: A Comment,” Journal of International Economics 14, 141-44.
Takayama, Akira (1982) “On Theorems of General Competitive Equilibrium of Production and
Trade: A Survey of Recent Developments in the Theory of International Trade,” Keio Economic
Studies 19, 1-38.
Thompson Henry (1985) “Complementarity in a Simple General Equilibrium Production Model,”
Canadian Journal of Economics 16, 616-21.
Thompson, Henry (1993) “The Magnification Effect with Three Factors, Keio Economic Studies 20,
57-64.
Thompson, Henry (2003) “Robustness of the Stolper-Samuelson Factor Intensity Price Link,” in
Handbook of International Trade, edited by Kwan Choi, Blackwell.
Thompson, Henry (2005) “Income Redistribution, Trade Prices, and International Capital in
Simulated Trade Models, in WTO and World Trade: Challenges in a New Era, edited by Geunter
Heiduk and Kar-yiu Wong, Springer-Verlag.
Thompson, Henry (2007) “An Empirical Measure of Factor Intensity when there are Many Factors
and Many Products,” The International Trade Journal 21, 109-19.
Walras, Léon (1874) Elements of Pure Economics, 1954 translation of 1926 edition, Richard Irwin.
13
Figure 1. Data series
Figure 2. Difference Stationary
Figure 3. Difference Stationary with 1975 Structural Break
14
Table 2. Spurious Regression
constant
lnL
3.95**
(2.35)
lnK
-0.69***
(-4.63)
lnE
0.25***
(3.36)
0.61***
(10.7)
lnpM
-0.01
(-0.10)
lnpS
EGτ -3.18
2
1.29***
(6.91)
adjR .996
DW 0.62 (+)
ARCH 2.64***
-3.32*
DW 1.46
ARCH 1.15
Table 3. Difference Model and ECM in ∆lnw
∆lnL
∆lnK
∆lnE
∆lnpM
∆lnpS
D
D∆lnL
D∆lnpM
Res-1
Difference
no
constant
-1.24***
(-3.10)
0.31***
(2.81)
0.74***
(8.14)
-0.37*
(-1.86)
0.49*
(1.81)
-0.01
(-0.50)
1.05
(1.20)
0.23
(0.81)
adjR2 .763
DW 1.39(?)
ARCH 1.87*
0.30***
(3.30)
-2.46***
(-4.74)
0.23**
(3.20)
0.65***
(7.53)
-0.21
(-1.09)
0.45*
(1.85)
-0.03**
(-2.30)
2.29**
(2.59)
0.09
(0.34)
adjR2 .738
DW 1.49
ARCH 0.93
ECM
no
constant
-1.22***
(-3.18)
0.26**
(2.45)
0.72***
(8.20)
-0.45**
(-2.31)
0.69**
(2.44)
-0.01
(-0.71)
1.09
(1.29)
0.29
(1.06)
-0.25**
(-2.28)
0.03***
(3.11)
-2.34***
(-4.64)
0.19*
(1.93)
0.64***
(7.62)
-0.28
(-1.53)
0.71*
(2.64)
-0.03**
(-2.36)
2.22**
(2.59)
0.15
(0.58)
-0.21**
(-2.04)
adjR2 .780
DW 1.37(?)
ARCH
2.43**
adjR2 .738
DW 1.49
ARCH 0.93
Table 4. Lagged ECM-1 in ∆lnw
∆lnL
∆lnK
∆lnE
∆lnpM
∆lnpS
D
D∆lnL
D∆lnpM
no
constant
-1.83***
(-3.56)
0.04
(0.26)
0.78***
(10.6)
0.13
(-0.75)
1.06***
(3.75)
-0.02*
(-1.69)
1.55**
(2.05)
0.01
(0.04)
0.01
(0.91)
-2.04***
(-3.62)
0.07
(0.53)
0.75***
(9.40)
-0.11
(-0.59)
0.97***
(3.20)
-0.03*
(-1.92)
1.82**
(2.24)
-0.02
(-0.07)
∆lnL-1
∆lnK-1
∆lnE-1
∆lnpM-1
∆lnpS-1
Res-1
Res-2
0.34
(0.70)
0.18*
(1.79)
0.14*
(1.69)
-0.14
(-0.86)
0.33
(1.34)
0.11
(0.83)
-0.53***
(-3.87)
0.20
(0.38)
0.14
(1.20)
0.12
(1.38)
-0.12
(-0.72)
0.26
(0.26)
0.11
(0.74)
-0.50***
(-3.48)
15
adjR2 .846
DW 2.02
ARCH 0.79
adjR2 .782
DW 2.11
ARCH 0.52
Table 5. Double Difference Model and ECM2 in ∆2lnw
∆2lnL
∆2lnK
∆2lnE
∆2lnpM
∆2lnpS
D
D∆2lnL
D∆2lnpM
-3.16***
(-3.51)
0.13
(1.09)
0.67***
(8.39)
0.00
(0.00)
0.64**
(2.32)
0.00
(0.27)
3.35**
(2.57)
-0.11
(-0.41)
0.00
(0.38)
-3.23***
(-3.49)
0.13
(1.07)
0.67***
(8.33)
0.00
(0.00)
0.65**
(2.33)
-0.00
(-0.12)
3.42**
(2.57)
-0.11
(-0.41)
ECM2
no
constant
0.00
(1.21)
-2.35***
(-3.09)
0.05
(0.50)
0.65***
(10.1)
-0.23
(-1.61)
0.83***
(3.63)
0.00
(0.27)
2.64**
(2.44)
0.31
(1.29)
-0.72***
(-4.91)
-2.48***
(-3.25)
0.04
(0.44)
0.66***
(10.2)
-0.24*
(-1.67)
0.89***
(3.81)
-0.00
(-0.76)
2.79**
(2.58)
0.32
(1.35)
-0.75***
(-5.05)
constant
Res∆
Table 6. Comparison of ECM Wage Elasticities
L / break
-1.39***
(6.78)
K
0.32***
(3.52)
E
0.87***
(17.8)
pM
-0.45**
(2.32)
pS
1.01***
(7.83)
ECM-1
-2.20*** / -0.65
(4.14) / (0.70)
0.32***
(2.76)
1.25***
(8.93)
0
1.74***
(5.01)
ECM2
-3.24*** / 0.59*
(-5.41) / (1.59)
0.22*
(1.33)
1.18***
(168)
-0.27***
(-4.43)
1.18***
(6.98)
average
-2.28 / -0.88
0.28
1.10
-0.36
1.31
ECM
16
View publication stats
adjR2 .667
DW 2.24
ARCH 0.80
adjR2 .682
DW 2.25
ARCH 1.01
adjR2 .768
DW 1.83
ARCH 0.82
adjR2 .791
DW 1.83
ARCH 1.12