Tax Policy and Local Labor Market Behavior: Online Appendix (Garrett, Ohrn and Suárez Serrato, May 2019)
Summary
The online appendix to "Tax Policy and Local Labor Market Behavior" by Daniel G. Garrett of Duke University, Eric Ohrn of Grinnell College and Juan Carlos Suárez Serrato of Duke University and NBER, dated May 2019. It states that the appendix includes additional information on the data and methods used in the paper as well as supplementary results. Appendix B contains variable definitions, including accelerated depreciation variables such as Duration and Long Duration Exposure. Appendix C describes additional results and robustness checks, Online Appendix D discusses the role of tax losses and Section 179 expensing rules, and Appendix E reports results for the effects of bonus depreciation on the employment-to-population ratio.
Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used
- Issuer
- Source documents
- Document type
- Tax Policy And Local Labor Market Behavior
- Case
- Tax Policy And Local Labor Market Behavior
Cite this document
“Tax Policy and Local Labor Market Behavior: Online Appendix (Garrett, Ohrn and Suárez Serrato, May 2019).”
Original at doi.org. Archived at Pandemic Darlings, https://pandemicdarlings.com/source-documents/tax-policy-and-local-labor-market-behavior/
, SHA-256 84ae86df73c8f874ab73bf70b5783d07683a8b3498fcdb68576116bc165aef61.Full text
Tax Policy and Local Labor Market Behavior
Daniel G. Garrett Eric Ohrn
Duke University Grinnell College
Juan Carlos Suárez Serrato
Duke University & NBER
May 2019
A Online Appendix: Not For Publication
This Online Appendix includes additional information on the data and methods used in the paper
as well as supplementary results. Appendix B contains additional details on our data sources.
Appendix C lists results from robustness checks that are mentioned in the body of the paper. Online
Appendix D discusses the role of tax losses and Section 179 expensing rules in the interpretation
of our results. Appendix E shows that we obtain similar results when we analyze the effects of
bonus depreciation on the employment-to-population ratio. Finally, additional tables and figures
are included in Appendix F.
1
B Variable Definitions
Variable name Definition
Accelerated Depreciation Variables
Industry We define the industry at the 4-digit NAICS level and 3-digit NAICS level in
different cases and denote the difference by mentioning the number of digits.
Employment Number of average workers listed in a geographic area and industrial grouping
in a given year according to ?, annual avg emplvl.
Duration The present value of depreciation deductions for the average asset in which
each industry invests from ?.
Long Duration Share of employment in each county in industries in the top tercile of indus-
Exposure tries as ranked by duration of average investment. This variable is always
normalized to the interquartile range (IQR).
Other Outcome Variables
Earnings Total payments made to workers in a geographic area and industrial grouping
in a given year according to ?, total annual wages.
Earnings-per- Total payments made to workers in a geographic area and industrial grouping
Worker in a given year divided by employment. From ?, this variable is created as
total annual wages divided by annual avg emplvl.
Other Control Variables
DPAD Share of employment in each county in industries in the top tercile of indus-
tries as ranked by Qualified Production Activities Income as a percent of sales
in 2005 derived from data compiled in ?.
Trade (China) County-level exposure to trade from China from ?.
Trade (NAFTA) County-level exposure to trade related to NAFTA from ?.
Routine Jobs County-level share of routine labor from ?.
Capital Stock Total capital stock, including structures, equipment, and intellectual property
products, in 2001 from ? allocated to counties using employment shares at the
3-digit NAICS level.
IP Stock Total intellectual property products in 2001 from ? allocated to counties using
employment shares at the 3-digit NAICS level.
Demographics County-level education outcomes including percent of population with col-
lege degrees and with less than a high school education as well as racial de-
mographics percent white and black from the American Community Survey.
Data compiled in ?.
2
C Additional Results
This appendix describes tables and figures that report additional details of the specifications in
Figure 2, as well as additional results.
• Descriptive Statistics. We include several figures and tables to more completely describe
the variation in duration both across space and across industries that we use as identifying
variation in exposure to accelerated depreciation.
– Figure F3 provides a summary description of the source of employment in long dura-
tion industries.
– A map of the geographic distribution of long duration industries without normalizing
within-state means to zero is shown in Figure F4.
– Table F1 describes the within sector variation in duration as well as shares of national
employment from ? and capital stock from ?. The final column shows total varia-
tion (coefficient of variation multiplied by employment weight) with the manufacturing
variation normalized to be equal to one.
– A list of the top and bottom ten counties with over 100,000 population in 2001 based
on the percent of their employment coming from long duration industries is shown in
Table F2.
– Table F3 shows additional county descriptive statistics associated with exposure to long
duration, population, and local capital stock.
• Robustness to Controls. We show the robustness of the county-level regressions of em-
ployment, earnings, and earnings-per-worker in a series of expanded results with different
controls and different definitions of key variables. The primary results displayed in Figure 2
are robust to the inclusion or exclusion of the controls.
– Figures F5, F6, and F7 show the robustness of the specifications in Figure 2 to taking
away the county-level controls or to removing the state-by-year fixed effects. Although
the 3-digit NAICS industry-by-year fixed effects are fundamental to the identification
3
strategy, the county controls and state-by-year fixed effects are included as extra con-
trols. We show that the additional controls do not have qualitative impacts on the main
results.
– Tables F4, F5, and F6 show the annual coefficients from Figure 2 for employment,
earnings, and earnings-per-worker, respectively. The tables all include five specifica-
tions where column (3) is the preferred specification with state-by-year and 3-digit
NAICS industry-by-year fixed effects as well as controls for county economics and
demographics.
• Robustness to Definition of Exposure and Placebo Tests. We also show the robustness of
the baseline results shown in Figure 2 to the definition of the long duration exposure at the
county level. Instead of defining firms in the top tercile of industries ranked by duration to
be “long” duration, we change the threshold to the top 25% and 40% of industries and show
that the results are unchanged. We also include a placebo with exposure to long duration
industries that primarily invest in structures and intellectual property, NAICS 2111, 4821,
5311, 7111, 7112, 7211, 7212, and all of 81, which are long duration industries with more
than five times more structures and IP than equipment.
– The choice to use the percentile instead of a continuous measure of average duration is
twofold: (1) the continuous measure is influenced by larger outliers in NAICS 22 that
invest in much longer duration assets than any other industry, so much of the variation
is driven by outliers and (2) the measurement of industry-level duration of investment
is a function of assumptions on discount rates that are avoided by discretizing the set of
long-duration industries. In order to check that the results are not sensitive to the deci-
sion to discretize bonus exposure, we define other cutoffs and examine the stability of
the coefficients of the other exposure measures. The analogue of Figure 2 is shown us-
ing exposure to the top 25% of long duration firms in Figure F8 and using the exposure
to the top 40% of long duration firms in Figure F9.
– Tables F7, F8, and F9, show the annual coefficients from Figure 2d for employment,
earnings, and earnings-per-worker, respectively. The labor market outcomes tables
each include five specifications where column (3) is the preferred specification with
4
state-by-year and 3-digit NAICS industry-by-year fixed effects as well as county eco-
nomic and demographic controls.
5
D Adjusting for Losses and Section 179
This appendix discusses the role of losses and Section 179 expensing rules in interpreting our
results. In particular, we clarify that the interpretation of our main result is that of an intent-to-
treat (ITT) effect. Our main estimate differs from the treatment on the treated (ATOT) for three
reasons. First, some companies may rely on Section 179 expensing instead of bonus. Second, some
companies may not take up the incentives of bonus depreciation if they plan to report tax losses.
A third complication is that bonus depreciation has varied in intensity across our time period. This
section clarifies the interpretation of our results in light of these three factors.
We make three related points in this appendix:
1. First, accounting for Section 179 has small effects on our reduced-form estimates of the ef-
fects of bonus depreciation. Specifically, our estimates would be 11% smaller in the absence
of Section 179 expensing.
2. Second, accounting for the fraction of firms with losses implies that the ATOT would be
33% larger than the ITT. Accounting for both losses and Section 179 results in estimates of
the ATOT that are 19% larger than our ITT estimates.
3. Finally, we show that the time-pattern of losses and Section 179 expensing limits has negli-
gible effects on the time-path of our reduced-form effects in Figure 2.
Marginal Investment Incentives with Losses and Section 179
As noted by ?, the effects of bonus depreciation interact with two important factors. The first is
corporate losses. Since firms can only get the immediate benefit from the bonus depreciation de-
duction if they owe corporate taxes, we would expect to find smaller effects when a larger fraction
of firms experience year-end losses. Second, Section 179 allows firms to fully expense capital
investments if the investment value is below a given threshold. A higher Section 179 limit could
therefore confound the effects of bonus.
In order to explore the role of these interactions, we start by characterizing the present dis-
counted value (PDV) of depreciation deductions. To do so, we make use of the following defini-
tions:
6
• Under the modified accelerated cost recovery system (MACRS), the PDV of depreciation
deductions for the marginal dollar is z0 .
• Under Bonus, the PDV of depreciation deductions for the marginal dollar is b + (1 − b)z0 .
Figure D1a shows how the policy parameter b varies over time. The average value of b over
our sample period is 39%.
• Under Section 179, the PDV of depreciation deductions for the marginal dollar is 1 if I j,t < I¯t ,
where I¯t is the Section 179 limit. Moreover, Share 179t = E[I[I j,t < I¯t ]]t is the share of
investment that is eligible for Section 179 expensing. Figure D1b reports data from ? that
describes the time variation in Share 179t . The Share 179t is relatively stable over out time
period with an average value that is close to 8%.
• Let I[Gains j,t ] be the event that a firm is in the gains domain and Share Gainst = E[I[Gains j,t ]]t .
Figure D1c uses data in corporate losses by industry from the IRS Statistics of Income and
describes the time variation in Share Gainst . Over our sample period, the average value of
Share Gainst is close to 75%.
For an individual firm j, the general value of depreciation deductions for the marginal dollar of
investment is:
z = (b + (1 − b)z0 ) × (1 − I[I j,t < I¯t ]) + 1 × I[I j,t < I¯t ]
= (b + (1 − b)z0 ) + I[I j,t < I¯t ][1 − b − (1 − b)z0 )]
= (b + (1 − b)z0 ) + I[I j,t < I¯t ][(1 − b)(1 − z0 )].
Taking the difference between this value and z0 we have :
z − z0 = (1 − z0 )b + I[I j,t < I¯t ][(1 − b)(1 − z0 )]
= (1 − z0 )[b + (1 − b)I[I j,t < I¯t ]].
Intuitively, Section 179 gives b = 1 when I j,t < I¯t so the combined policy of bonus and Section 179
has a larger effect on z − z0 whenever the event I[I j,t < I¯t ] is more likely.
7
Assume now that a firm only values depreciation deductions in the gains domain. The average
value of the shock in a county is then:
E[z − z0 ]c,t = (1 − z0 ) × Share Gainst × [b + (1 − b)Share 179t ]
≈ Exposurec × Share Gainst × [b + (1 − b)Share 179t ], (1)
where we use our Exposurec measure as the empirical approximation of (1 − z0 ) .
Adjusting Average Reduced-Form Effects for Losses and Section 179
Equation 1 formalizes the notion that estimates that rely on Exposurec for identifying variation
will result in estimates of intent-to-treat effects. To see this, assume average values of b = 39%
and Share Gainst = 75% and temporarily ignore the role of Section 179 by setting Share 179t = 0.
Equation 1 then suggests that to recover the ATOT we would need to divide our estimates by
1
Share Gainst = 75%, which would increase their magnitude by 33% (≈ 0.75 ).
To understand the role of Share 179t , assume that Share Gainst = 1 and b = 39%. To obtain
the equivalent effect of an average bonus rate of b = 39% absent Section 179, we would need
39%
to multiply our estimates by: 39%+(1−39%)×8% ≈ 0.89, which would make them 11% smaller.
For instance, column (2) in Table 1 shows that the average increase in employment from an IQR
increase in exposure to bonus depreciation was 1.9%. Accounting for the role of Section 179, our
estimate would be 1.7% = 1.9% × 0.89.
To offset the effects of both losses and Section 179, we would have to multiply our estimates by
39%
75%×[39%+(1−39%)×8%] ≈ 1.19. The combined effect of losses and Section 179 would be to make
our estimates 19% larger. Absent Section 179 and in a world where no firms were constrained
in claiming bonus due to loss effects, we would expect to find an increase in employment of
2.3% = 1.9% × 1.19.
Similarly, suppose that we are interested in evaluating the effects of a policy where b = 50%
for a decade. Again, assuming no Section 179 and no frictions from corporate losses, we would
50%
expect an increase in employment of 2.9% = 1.9% × 1.52 where 1.52 = 75%×[39%+(1−39%)×8%] .
Adjusting Dynamics of Reduced-Form Effects for Losses and Section 179
As discussed above, while corporate losses and Section 179 expensing interact with bonus de-
preciation, accounting for these interactions has small effects on the interpretation of our average
8
estimates. An additional concern is that the time patterns in b, Share Gainst , and Share 179t influ-
ence the dynamics of the effects shown in Figure 2. We now perform similar adjustments as above
to show that this is not the case.
Conceptually, Equation 1 shows that our treatment is time-varying, and that the intensity of the
policy depends on the time patterns of b, Share Gainst , and Share 179t . The goal of this exercise is
to use our estimates and the time patterns in b, Share Gainst , and Share 179t from Figures D1a-D1c
to compare the observed policy to a counterfactual policy where b, Share Gainst , and Share 179t
are held constant at their average values over our time period.
To do so, Figure D1d plots the value of the adjustment factor Share Gainst ×[b+(1−b)Share 179t ]
over time. This plot mostly follows the time path of b; however, the amplitude of the curve is di-
minished by Share Gainst and the minimum value is augmented by Share 179t . Because outcomes
in a given year t are affected by the policy in previous years, we adjust our estimates by the cumu-
lative average of Share Gainst × [b + (1 − b)Share 179t ] from 2001 until a given year t. Figure D1e
plots this cumulative average relative to the average value of Share Gainst ×[b+(1−b)Share 179t ]
over the time period. We can then divide our estimates in Figure 2 by the values of Figure D1e
to obtain the reduced-form effects of a policy where b, Share Gainst , and Share 179t are held con-
stant at their average values over our time period. Figure D1d shows that a time-consistent policy
would result in larger effects in years 2002 and 2006-2009. Similarly, this adjustment would imply
smaller effects in years 2003-2004 and 2011-2012.
Figure D1f shows that adjusting our estimates on the effects of bonus depreciation on employ-
ment so that they have a time-consistent interpretation results in very similar effects. The largest
change is that we observe a slightly larger effect in years 2006-2007.
Overall, the pattern of losses and Section 179 expensing do not play a material role in explain-
ing the dynamics of how bonus depreciation affects the labor market. For this reason, we present
the un-adjusted results in the paper. This result is also consistent with results in ? that show that
business investment was similarly responsive to bonus depreciation in the early and latter years of
our sample.
9
Figure D1: Adjusting for Losses and Section 179 in the Employment Effects of Bonus Depre-
ciation
A. Bonus Depreciation Rate b B. Fraction of §179-Eligible Qualified Investment
1 .1
.8
Percent Percentage Share
.4 .6
.06 .08
.2
0 .04
2002 2004 2006 2008 2010 2012 2002 2004 2006 2008 2010 2012
Year Year
Bonus Depreciation Rate Qualified Investment Eligible for S179
C. Fraction of Firms with Gains D. Adjustment Factor:
Share Gainst × [b + (1 − b)Share 179t ]
.85 .8
.8
Percentage Share
Percentage Share
.7 .75 .4 .6
.2
.65
.6 0
1998 2000 2002 2004 2006 2008 2010 2012 2002 2004 2006 2008 2010 2012
Year Year
Assets in Firms with Gains Revenues in Firms with Gains Treatment Portion (Assets) Treatment Portion (Revenues)
E. Normalized Cumulative Average of F. Adjusted Effect of Bonus Depreciation
Adjustment Factor on Employment
1.4 .06
Employment
Percentage Share
-.03 0 .03
1 1.2
-.06
1997 2000 2003 2006 2009 2012
Year
.8
Baseline 95% CI
Loss and S179 Adjusted (Assets) 95% CI
2002 2004 2006 2008 2010 2012 Loss and S179 Adjusted (Sales) 95% CI
Year
Recession 30% Bonus
Treatment Portion (Assets) Treatment Portion (Revenues) 50% Bonus 100% Bonus
Notes: Author’s calculations using employment data from QCEW, industry duration data from ?, net operating loss
shares from ?, and Section 179 use from authors calculations and results reported in ?. This figure shows the annual
coefficients from an event study around the implementation of bonus depreciation that is adjusted for national net
operating losses and access to Section 179. Section D discusses a correction to our baseline estimates that adjusts for
the intensity of treatment from Bonus in a given year due to interactions with losses and Section 179. Figure D1a
shows the bonus rate b for each year. Figure D1b shows the fraction of total investment that is eligible for Section
179 deductions. Figure D1c shows the percent of assets and revenues in firms that do not have losses. Figures D1a,
D1b, and D1c combine into Figure D1d, the adjustment factor, and Figure D1e, the normalized cumulative average
adjustment factor. Dividing the regression results from Figure 2a by the adjustment factor yields the adjusted effect of
Bonus on Employment shown in Figure D1f.
10
E Effects of Bonus Depreciation on the Employment-to-Population
Ratio
One potential mechanism behind the increase in employment is the geographic relocation of work-
ers. In order to account for this factor, we estimate the effects of our shock on the employment-to-
population ratio, as in ?.
Figure E2 plots the results of this analysis and shows that, similar to Figure 2, the effects of
bonus depreciation on employment crest in 2006.1 This figure shows that, by 2006, a unit IQR
increase in Exposure increases the employment-to-population ratio by 1 percentage point.
Overall, the average effect for years 2003-2012 is that a unit IQR increase in Exposure raised
the employment-to-population ratio by 0.76 percentage points. Relative to the average US working-
age population during our period of 195 million, this implies that the average effect of Exposure
would be to raise employment by 4.06 million jobs. Comparing this employment effect with the
cost of the policy implies a cost-per-job of $73,000 ≈ 297.5
4.06 .
Our discussion in the paper focuses on the cost of creating a job in a given location. For this
reason, our main estimate of $53,000 is smaller than the estimate of $73,000, which applies to the
cost of creating a job relative to a given population. We choose to focus on the percentage change
in employment since this outcome is comparable to previous work on local fiscal multipliers.
Finally, it is worth noting that the dynamics of the effects of bonus depreciation in Figure E2
are similar those of our main result in Figure 2. Specifically, bonus depreciation has temporary
effects on employment. While these level effects are persistent, bonus depreciation does not lead
to sustained increases in the rate of employment growth.
1 These regressions use the county-year outcome as the unit of observations, which does not allow us to control
for industry by year fixed effects. For details regarding this specification, see Appendix F of an earlier version of this
article in NBER Working Paper 25546.
11
Figure E2: Effects of Bonus Depreciation on Employment-to-Population Ratio
2
Employment Population Ratio
-.5 0 .5 1 1.5
-1
1997 2000 2003 2006 2009 2012
Year
Long Duration Exposure 95% CI
Recession 30% Bonus
50% Bonus 100% Bonus
Note: Author’s calculations using data from QCEW and ?. This figure shows the annual coefficients from an event
study around the implementation of bonus depreciation. The dependent variable is the change in in the Employment-
to-Population ratio. The variable of interest is the percent of employment that is resides in long duration industries
normalized to the interquartile range (IQR). For more information on this calculation, see Appendix E. Standard errors
are clustered at the county level.
12
F Additional Figures and Tables
13
Figure F3: Percent of Long Duration Employment Derived from Each 2-digit NAICS
Accommodation and food services
Administrative and waste management services
Agriculture, forestry, fishing, and hunting
Arts, entertainment, and recreation
Construction
Educational services
Finance and insurance
Health care and social assistance
Information
Management of companies and enterprises
Manufacturing
Mining
Other services, except government
Professional, scientific, and technical services
Real estate and rental and leasing
Retail trade
Transportation and warehousing
Utilities
Wholesale trade
0 .1 .2 .3 .4
Percent of Long Duration Employment
Note: Author’s calculations using data from QCEW and ?. This figure shows the percent of long duration employment
coming from each sector as defined by 2-digit NAICS in 2001. Data are at the national level.
Figure F4: Exposure to Long Duration Industries in 2001, Raw
0.27 − 0.82
0.22 − 0.27
0.17 − 0.22
0.11 − 0.17
0.00 − 0.11
No data
Note: Author’s calculations using data from QCEW and ?. This figure shows the percent of employment in each
county that comes from the top three deciles of employment-weighted industries by average duration of investment.
Industries are defined by 4-digit NAICS codes. A version of this map normalized to standard deviations from state-
level mean is shown in Figure 1b.
14
Figure F5: Employment Effects of Bonus Depreciation, Controls Robustness
.06
Employment
0 .03
-.03
1997 2000 2003 2006 2009 2012
Year
3-digit NAICS-by-Year FE 95% CI
3-digit NAICS- and State-by-Year FE 95% CI
FE and County Controls 95% CI
Recession 30% Bonus
50% Bonus 100% Bonus
Note: Author’s calculations using data from QCEW and ?. This figure shows the annual coefficients from an event
study around the implementation of bonus depreciation. The dependent variable in Figure F5 is employment relative
to 2001. The covariate of interest is exposure to long duration, as in Figure 2a. The coefficients and additional
specifications for Figure F5 are shown in Table F4. See Appendix C for additional discussion about the importance of
controls. Standard errors are clustered at the county level.
15
Figure F6: Earnings Effects of Bonus Depreciation, Controls Robustness
.06
Earnings
.03
0
-.03
1997 2000 2003 2006 2009 2012
Year
3-digit NAICS-by-Year FE 95% CI
3-digit NAICS- and State-by-Year FE 95% CI
FE and County Controls 95% CI
Recession 30% Bonus
50% Bonus 100% Bonus
Note: Author’s calculations using data from QCEW and ?. This figure shows the annual coefficients from an event
study around the implementation of bonus depreciation. The dependent variable in Figure F6 is earnings relative
to 2001. The covariate of interest is exposure to long duration, as in Figure 2b. The coefficients and additional
specifications for Figure F6 are shown in Table F5. See Appendix C for additional discussion about the importance of
controls. Standard errors are clustered at the county level.
16
Figure F7: Earnings-per-Worker Effects of Bonus Depreciation, Controls Robustness
.06
Earnings-per-Worker
0 .03
-.03
1997 2000 2003 2006 2009 2012
Year
3-digit NAICS-by-Year FE 95% CI
3-digit NAICS- and State-by-Year FE 95% CI
FE and County Controls 95% CI
Recession 30% Bonus
50% Bonus 100% Bonus
Note: Author’s calculations using data from QCEW and ?. This figure shows the annual coefficients from an event
study around the implementation of bonus depreciation. The dependent variable in Figure F7 is earnings-per-worker
relative to 2001. The covariate of interest is exposure to long duration, as in Figure 2c. The coefficients and additional
specifications for Figure F7 are shown in Table F6. See Appendix C for additional discussion about the importance of
controls. Standard errors are clustered at the county level.
17
Figure F8: Effects of Bonus Depreciation by Exposure to Long Duration Industries, Long
Duration Cutoff at 25%
A. Employment B. Earnings
.06 .06
Employment
.03
Earnings
0 .03
0
-.03 -.03
1997 2000 2003 2006 2009 2012 1997 2000 2003 2006 2009 2012
Year Year
Long Duration Exposure 95% CI Long Duration Exposure 95% CI
Recession 30% Bonus Recession 30% Bonus
50% Bonus 100% Bonus 50% Bonus 100% Bonus
C. Earnings-per-Worker
.06
Earnings-per-Worker
0 .03
-.03
1997 2000 2003 2006 2009 2012
Year
Long Duration Exposure 95% CI
Recession 30% Bonus
50% Bonus 100% Bonus
Note: Author’s calculations using data from QCEW and ?. This figure shows the annual coefficients from an event
study around the implementation of bonus depreciation. The dependent variable is Employment in Figure F8a, Earn-
ings in Figure F8b and Earnings-per-Worker in Figure F8c. The variable of interest is the percent of employment
that is resides in long duration industries normalized to the interquartile range (IQR). These regressions correspond to
those displayed in Figure 2 with long duration industries defined at the 25% cutoff instead of 33%.
18
Figure F9: Effects of Bonus Depreciation by Exposure to Long Duration Industries, Long
Duration Cutoff at 40%
A. Employment B. Earnings
.06 .06
Employment
.03
Earnings
0 .03
0
-.03 -.03
1997 2000 2003 2006 2009 2012 1997 2000 2003 2006 2009 2012
Year Year
Long Duration Exposure 95% CI Long Duration Exposure 95% CI
Recession 30% Bonus Recession 30% Bonus
50% Bonus 100% Bonus 50% Bonus 100% Bonus
C. Earnings-per-Worker
.06
Earnings-per-Worker
0 .03
-.03
1997 2000 2003 2006 2009 2012
Year
Long Duration Exposure 95% CI
Recession 30% Bonus
50% Bonus 100% Bonus
Note: Author’s calculations using data from QCEW and ?. This figure shows the annual coefficients from an event
study around the implementation of bonus depreciation. The dependent variable is Employment in Figure F9a, Earn-
ings in Figure F9b and Earnings-per-Worker in Figure F9c. The variable of interest is the percent of employment
that is resides in long duration industries normalized to the interquartile range (IQR). These regressions correspond to
those displayed in Figure 2 with long duration industries defined at the 40% cutoff instead of 33%.
19
Table F1: Characteristics of Investment Duration by Sector
NAICS Industry Average SD CV Employment Capital Variation
11 Agriculture, forestry, fishing, and hunting 0.862 0.010 1.160 0.945 3.658 0.116
21 Mining 0.881 0.008 0.940 0.382 2.353 0.038
22 Utilities 0.767 0.032 4.110 0.183 6.613 0.079
23 Construction 0.894 0.003 0.320 7.300 3.183 0.247
31-33 Manufacturing 0.880 0.008 0.870 10.892 24.917 1.000
42 Wholesale trade 0.888 0.004 0.450 5.295 4.875 0.251
44-45 Retail trade 0.881 0.009 1.010 15.518 3.929 1.654
48-49 Transportation and warehousing 0.890 0.016 1.830 3.633 9.191 0.702
51 Information 0.879 0.018 2.070 3.033 10.031 0.663
20 52 Finance and insurance 0.887 0.006 0.700 4.387 8.860 0.324
53 Real estate and rental and leasing 0.878 0.019 2.180 2.157 7.638 0.496
54 Professional, scientific, and technical services 0.893 0.003 0.300 7.489 2.527 0.237
55 Management of companies and enterprises 0.880 . . 1.811 0.928 .
56 Administrative and waste management services 0.892 0.003 0.280 8.157 1.639 0.241
61 Educational services 0.885 . 0.000 1.663 0.559 0.000
62 Health care and social assistance 0.880 0.009 1.070 11.038 4.794 1.246
71 Arts, entertainment, and recreation 0.858 0.015 1.800 1.393 0.767 0.265
72 Accommodation and food services 0.870 0.004 0.460 10.750 2.148 0.522
81 Other services, except government 0.876 0.011 1.250 3.974 1.389 0.524
Note: Author’s calculations using data from QCEW and ?. This table shows the average duration characteristics of each 2-digit NAICS sector. The Variation column
shows within-sector variation, defined as the coefficient of variation multiplied by the employment weight, relative to manufacturing in 2001. Sector variables are
calculated by aggregating data at the industry level using employment shares from QCEW.
Table F2: List of Counties by Exposure to Long Duration Industries
Rank County Long Duration Employment Exposure
1 Kent County, Delaware .1132373
2 Durham County, North Carolina .1204463
3 Forsyth County, Georgia .1347249
4 Sullivan County, Tennessee .1513936
5 Olmsted County, Minnesota .1519782
6 Catawba County, North Carolina .1568298
7 Rankin County, Mississippi .1573195
8 Sarpy County, Nebraska .1583619
9 New Castle County, Delaware .1604007
10 Clayton County, Georgia .161008
508 Clark County, Nevada .4558978
509 Cape May County, New Jersey .4573346
510 Merced County, California .4581397
511 Napa County, California .4653518
512 Fresno County, California .471666
513 Yuma County, Arizona .4897471
514 Monterey County, California .4994023
515 Yakima County, Washington .5202556
516 Tulare County, California .5315269
517 Atlantic County, New Jersey .5559594
Note: Author’s calculations using data from QCEW and ?. This table lists the top ten and bottom ten major coun-
ties based on their exposure to long duration industries. This list only includes counties with more than 100,000 in
population in 2000.
21
Table F3: County Level Descriptive Statistics
Mean SD 25th 50th 75th
Total Population, 2001 92217.745 298364.878 11596.000 25449.000 63130.000
Total Employment 34587.489 126911.918 2338.000 6426.000 19019.000
Employment Growth, 2001-2007 0.053 0.185 -0.040 0.038 0.122
Employment Growth, 2001-2012 0.025 0.269 -0.101 -0.005 0.107
Number of 3-Digit NAICS Industries 36.540 20.004 20.000 35.000 50.000
County Capital
22 Equipment Stock, 2001 1226.515 5151.298 43.399 165.312 576.610
Intellectual Property Stock, 2001 430.718 2053.905 4.027 22.121 133.989
Exposure to Bonus Depreciation
Average NPV of Depreciation (No Bonus) 0.879 0.005 0.877 0.879 0.882
Long Duration Exposure 0.206 0.096 0.142 0.203 0.259
Long Duration Exposure, 25% 0.168 0.087 0.111 0.160 0.210
Long Duration Exposure, 40% 0.256 0.113 0.178 0.257 0.331
Exposure to Non-Eligible Long Duration 0.021 0.036 0.003 0.012 0.026
Note: Author’s calculations using data from QCEW and ?. This table displays descriptive characteristics of the county level exposure to long duration industries.
The Mean column displays the mean across counties and the SD column displays the standard deviation. The following three columns display the 25th , 50th , and
75th percentile of the distribution, respectively.
Table F4: Event Study Regression of Total Employment on Exposure to Long Duration In-
dustries
Exposure to Long Duration Industries (1) (2) (3) (4) (5)
X 1997 -0.003 0.003 -0.004 -0.003 0.000
(0.005) (0.004) (0.003) (0.003) (0.004)
X 1998 -0.004 0.001 -0.003 -0.004 0.001
(0.004) (0.003) (0.003) (0.003) (0.004)
X 1999 -0.004 0.001 -0.001 -0.003 0.001
(0.003) (0.002) (0.002) (0.002) (0.003)
X 2000 -0.002 -0.001 -0.001 -0.001 0.000
(0.002) (0.002) (0.002) (0.002) (0.002)
X 2002 0.010∗∗∗ 0.010∗∗∗ 0.007∗∗∗ 0.004∗∗∗ 0.008∗∗∗
(0.002) (0.002) (0.002) (0.002) (0.002)
X 2003 0.015∗∗∗ 0.017∗∗∗ 0.013∗∗∗ 0.010∗∗∗ 0.013∗∗∗
(0.003) (0.004) (0.003) (0.002) (0.004)
X 2004 0.023∗∗∗ 0.024∗∗∗ 0.020∗∗∗ 0.016∗∗∗ 0.020∗∗∗
(0.005) (0.005) (0.004) (0.003) (0.005)
X 2005 0.028∗∗∗ 0.027∗∗∗ 0.024∗∗∗ 0.020∗∗∗ 0.026∗∗∗
(0.006) (0.006) (0.005) (0.004) (0.006)
X 2006 0.028∗∗∗ 0.027∗∗∗ 0.022∗∗∗ 0.018∗∗∗ 0.024∗∗∗
(0.008) (0.007) (0.006) (0.005) (0.008)
X 2007 0.024∗∗∗ 0.024∗∗∗ 0.019∗∗∗ 0.016∗∗∗ 0.021∗∗∗
(0.009) (0.007) (0.006) (0.006) (0.008)
X 2008 0.019∗∗ 0.020∗∗∗ 0.018∗∗∗ 0.016∗∗∗ 0.019∗∗
(0.008) (0.006) (0.006) (0.006) (0.008)
X 2009 0.012 0.017∗∗∗ 0.016∗∗∗ 0.013∗∗ 0.017∗∗
(0.008) (0.006) (0.005) (0.005) (0.008)
X 2010 0.011 0.017∗∗∗ 0.019∗∗∗ 0.016∗∗∗ 0.021∗∗∗
(0.007) (0.006) (0.006) (0.005) (0.008)
X 2011 0.010 0.014∗∗ 0.019∗∗∗ 0.017∗∗∗ 0.021∗∗
(0.008) (0.006) (0.006) (0.006) (0.008)
X 2012 0.011 0.012∗ 0.021∗∗∗ 0.019∗∗∗ 0.025∗∗∗
(0.008) (0.007) (0.006) (0.006) (0.009)
3-digit Industry-by-Year Fixed Effects Yes Yes Yes Yes Yes
State-by-Year Fixed Effects Yes Yes Yes Yes
County Characteristics Yes Yes Yes
Winsorized Weights Yes
Drops Small County-Sectors (<1000) Yes
Note: Author’s calculations using data from QCEW and ?. The coefficients displayed in this table come from an
event study regression of total employment. The dependent variable is the percent change in total Earnings relative to
2001. The variable of interest is the percent of employment that is derived from long duration industries normalized
to the interquartile range (IQR). Standard errors clustered at the county level are shown in parentheses. Column (3) is
corresponds to Figure 2a.
∗ p < 0.10, ∗∗ p < 0.05, ∗∗∗ p < 0.01.
23
Table F5: Event Study Regression of Total Earnings on Exposure to Long Duration Industries
Exposure to Long Duration Industries (1) (2) (3) (4) (5)
X 1997 0.004 0.010∗∗∗ -0.002 -0.002 0.001
(0.004) (0.003) (0.003) (0.003) (0.004)
X 1998 0.003 0.007∗∗ -0.002 -0.002 0.002
(0.003) (0.003) (0.003) (0.002) (0.003)
X 1999 0.000 0.003∗ -0.001 -0.001 0.001
(0.002) (0.002) (0.002) (0.002) (0.003)
X 2000 -0.004 -0.005 -0.004 -0.003 -0.005
(0.002) (0.003) (0.003) (0.002) (0.004)
X 2002 0.013∗∗∗ 0.015∗∗∗ 0.009∗∗∗ 0.007∗∗∗ 0.010∗∗∗
(0.003) (0.003) (0.002) (0.002) (0.003)
X 2003 0.019∗∗∗ 0.022∗∗∗ 0.016∗∗∗ 0.012∗∗∗ 0.016∗∗∗
(0.004) (0.005) (0.004) (0.003) (0.005)
X 2004 0.028∗∗∗ 0.031∗∗∗ 0.024∗∗∗ 0.019∗∗∗ 0.024∗∗∗
(0.006) (0.006) (0.005) (0.004) (0.006)
X 2005 0.035∗∗∗ 0.035∗∗∗ 0.029∗∗∗ 0.025∗∗∗ 0.033∗∗∗
(0.009) (0.007) (0.006) (0.006) (0.008)
X 2006 0.035∗∗∗ 0.034∗∗∗ 0.026∗∗∗ 0.022∗∗∗ 0.029∗∗∗
(0.011) (0.009) (0.007) (0.007) (0.010)
X 2007 0.030∗∗ 0.030∗∗∗ 0.020∗∗∗ 0.018∗∗ 0.026∗∗
(0.013) (0.009) (0.008) (0.007) (0.010)
X 2008 0.026∗∗ 0.029∗∗∗ 0.017∗∗ 0.015∗∗ 0.021∗
(0.013) (0.009) (0.008) (0.007) (0.011)
X 2009 0.020∗ 0.028∗∗∗ 0.014∗ 0.011 0.017
(0.012) (0.008) (0.007) (0.007) (0.011)
X 2010 0.013 0.023∗∗∗ 0.012 0.012 0.016
(0.012) (0.008) (0.008) (0.007) (0.011)
X 2011 0.011 0.017∗ 0.009 0.009 0.013
(0.013) (0.009) (0.008) (0.008) (0.012)
X 2012 0.011 0.011 0.008 0.010 0.017
(0.014) (0.011) (0.009) (0.009) (0.013)
3-digit Industry-by-Year Fixed Effects Yes Yes Yes Yes Yes
State-by-Year Fixed Effects Yes Yes Yes Yes
County Characteristics Yes Yes Yes
Winsorized Weights Yes
Drops Small County-Sectors (<1000) Yes
Note: Author’s calculations using data from QCEW and ?. The coefficients displayed in this table come from an event
study regression of total Earnings. The dependent variable is the percent change in total earnings relative to 2001.
The variable of interest is the percent of employment that is derived from long duration industries normalized to the
interquartile range (IQR). Standard errors clustered at the county level are shown in parentheses. Column (3) is shown
as Figure 2b.
∗ p < 0.10, ∗∗ p < 0.05, ∗∗∗ p < 0.01.
24
Table F6: Event Study Regression of Earnings-per-Worker on Exposure to Long Duration
Industries
Exposure to Long Duration Industries (1) (2) (3) (4) (5)
X 1997 0.006∗∗∗ 0.008∗∗∗ 0.001 0.001 -0.000
(0.002) (0.002) (0.001) (0.001) (0.002)
X 1998 0.006∗∗∗ 0.006∗∗∗ 0.001 0.001 0.001
(0.002) (0.002) (0.002) (0.001) (0.002)
X 1999 0.003∗∗ 0.002 -0.001 0.001 -0.001
(0.001) (0.001) (0.001) (0.001) (0.001)
X 2000 -0.001 -0.004∗∗ -0.004∗∗ -0.002∗ -0.006∗∗
(0.001) (0.002) (0.002) (0.001) (0.002)
X 2002 0.003∗∗∗ 0.004∗∗∗ 0.003∗∗∗ 0.002∗∗∗ 0.003∗∗
(0.001) (0.001) (0.001) (0.001) (0.001)
X 2003 0.003∗∗ 0.005∗∗∗ 0.002∗ 0.002 0.003∗
(0.001) (0.001) (0.001) (0.001) (0.002)
X 2004 0.003∗∗ 0.004∗∗∗ 0.002 0.002 0.002
(0.002) (0.001) (0.001) (0.001) (0.002)
X 2005 0.003 0.004∗∗∗ 0.002 0.002 0.003
(0.002) (0.002) (0.002) (0.002) (0.002)
X 2006 0.001 0.003 -0.000 0.001 0.002
(0.003) (0.002) (0.002) (0.002) (0.003)
X 2007 -0.000 0.002 -0.002 0.000 0.000
(0.004) (0.003) (0.002) (0.002) (0.003)
X 2008 0.002 0.005 -0.005∗ -0.003 -0.003
(0.004) (0.003) (0.002) (0.002) (0.003)
X 2009 0.004 0.008∗∗∗ -0.006∗∗ -0.005∗∗ -0.004
(0.004) (0.003) (0.002) (0.002) (0.003)
X 2010 -0.003 0.001 -0.013∗∗∗ -0.009∗∗∗ -0.012∗∗∗
(0.005) (0.003) (0.003) (0.002) (0.004)
X 2011 -0.003 -0.001 -0.016∗∗∗ -0.012∗∗∗ -0.016∗∗∗
(0.005) (0.004) (0.003) (0.003) (0.004)
X 2012 -0.005 -0.004 -0.018∗∗∗ -0.013∗∗∗ -0.017∗∗∗
(0.005) (0.004) (0.003) (0.003) (0.004)
3-digit Industry-by-Year Fixed Effects Yes Yes Yes Yes Yes
State-by-Year Fixed Effects Yes Yes Yes Yes
County Characteristics Yes Yes Yes
Winsorized Weights Yes
Drops Small County-Sectors (<1000) Yes
Note: Author’s calculations using data from QCEW and ?. The coefficients displayed in this table come from an event
study regression of earnings-per-worker. The dependent variable is the percent change in earnings-per-worker relative
to 2001. The variable of interest is the percent of employment that is derived from long duration industries normalized
to the interquartile range (IQR). Standard errors clustered at the county level are shown in parentheses. Column (3) is
shown as Figure 2c.
∗ p < 0.10, ∗∗ p < 0.05, ∗∗∗ p < 0.01.
25
Table F7: Event Study Regression of Employment on Exposure to Structures Intensive Long
Duration Industries (Placebo Test)
Exposure to Placebo Industries (1) (2) (3) (4) (5)
X 1997 -0.009∗∗∗ -0.002 -0.002 -0.002 -0.002
(0.003) (0.002) (0.002) (0.002) (0.002)
X 1998 -0.007∗∗ -0.001 -0.001 -0.001 -0.000
(0.003) (0.002) (0.002) (0.002) (0.002)
X 1999 -0.005∗∗∗ -0.001 -0.000 -0.000 0.000
(0.002) (0.001) (0.001) (0.001) (0.001)
X 2000 -0.002∗ 0.001 0.001 0.000 0.001
(0.001) (0.001) (0.001) (0.001) (0.001)
X 2002 0.001 -0.000 -0.000 -0.000 -0.000
(0.001) (0.001) (0.001) (0.001) (0.001)
X 2003 0.004∗ 0.000 0.001 0.000 0.000
(0.002) (0.001) (0.001) (0.001) (0.001)
X 2004 0.009∗∗ 0.001 0.002 0.002 0.001
(0.004) (0.002) (0.002) (0.002) (0.002)
X 2005 0.013∗∗ 0.001 0.002 0.002 0.001
(0.005) (0.003) (0.002) (0.003) (0.003)
X 2006 0.016∗∗ 0.001 0.001 0.001 0.000
(0.007) (0.004) (0.004) (0.004) (0.005)
X 2007 0.016∗∗ 0.001 -0.001 0.000 -0.002
(0.007) (0.004) (0.004) (0.004) (0.005)
X 2008 0.015∗∗ 0.000 -0.001 -0.000 -0.002
(0.007) (0.004) (0.004) (0.004) (0.005)
X 2009 0.011∗∗ -0.001 -0.002 -0.002 -0.003
(0.006) (0.004) (0.004) (0.004) (0.005)
X 2010 0.010∗ -0.001 -0.002 -0.001 -0.003
(0.005) (0.004) (0.004) (0.004) (0.005)
X 2011 0.010∗ -0.002 -0.002 -0.001 -0.003
(0.005) (0.004) (0.004) (0.005) (0.005)
X 2012 0.010∗ -0.001 -0.002 -0.000 -0.003
(0.005) (0.004) (0.004) (0.005) (0.005)
3-digit Industry-by-Year Fixed Effects Yes Yes Yes Yes Yes
State-by-Year Fixed Effects Yes Yes Yes Yes
County Characteristics Yes Yes Yes
Winsorized Weights Yes
Drops Small County-Sectors (<1000) Yes
Note: Author’s calculations using data from QCEW and ?. The coefficients displayed in this table come from an event
study regression of employment. The dependent variable is the percent change in employment relative to 2001. The
variable of interest is the percent of employment that is derived from long duration industries with more than five times
more structures and intellectual property than equipment normalized to the interquartile range (IQR). Standard errors
clustered at the county level are shown in parentheses. Column (3) is shown in Figure 2d.
∗ p < 0.10, ∗∗ p < 0.05, ∗∗∗ p < 0.01.
26
Table F8: Event Study Regression of Total Earnings on Exposure to Structures Intensive
Long Duration Industries (Placebo Test)
Exposure to Placebo Industries (1) (2) (3) (4) (5)
X 1997 0.001 0.001 0.000 0.000 0.000
(0.001) (0.001) (0.001) (0.001) (0.001)
X 1998 0.001 -0.001 -0.001 -0.001 -0.001
(0.001) (0.001) (0.001) (0.001) (0.001)
X 1999 0.001 -0.001 -0.001 -0.001 -0.001
(0.001) (0.001) (0.001) (0.000) (0.001)
X 2000 0.000 -0.000 -0.000 -0.000 -0.001
(0.001) (0.001) (0.001) (0.000) (0.001)
X 2002 0.001 0.000 0.001 0.001∗ 0.001
(0.000) (0.000) (0.001) (0.000) (0.001)
X 2003 0.001 -0.000 -0.000 -0.000 -0.000
(0.001) (0.001) (0.001) (0.001) (0.001)
X 2004 0.002∗∗ 0.000 0.000 0.000 0.000
(0.001) (0.001) (0.001) (0.001) (0.001)
X 2005 0.004∗∗∗ 0.002∗∗ 0.002∗ 0.002∗∗ 0.002
(0.001) (0.001) (0.001) (0.001) (0.001)
X 2006 0.005∗∗∗ 0.003∗∗ 0.002 0.002 0.003
(0.001) (0.001) (0.001) (0.001) (0.002)
X 2007 0.006∗∗∗ 0.004∗∗ 0.002 0.001 0.002
(0.002) (0.002) (0.001) (0.001) (0.002)
X 2008 0.006∗∗∗ 0.004∗∗ 0.001 0.001 0.002
(0.002) (0.002) (0.002) (0.001) (0.002)
X 2009 0.006∗∗∗ 0.003∗∗ -0.000 -0.001 0.001
(0.002) (0.001) (0.001) (0.001) (0.002)
X 2010 0.005∗∗∗ 0.004∗∗ -0.001 -0.001 0.000
(0.002) (0.002) (0.002) (0.001) (0.002)
X 2011 0.005∗∗ 0.004∗ -0.001 -0.001 -0.000
(0.002) (0.002) (0.002) (0.001) (0.002)
X 2012 0.003 0.002 -0.003∗ -0.002 -0.003
(0.002) (0.002) (0.002) (0.002) (0.003)
3-digit Industry-by-Year Fixed Effects Yes Yes Yes Yes Yes
State-by-Year Fixed Effects Yes Yes Yes Yes
County Characteristics Yes Yes Yes
Winsorized Weights Yes
Drops Small County-Sectors (<1000) Yes
Note: Author’s calculations using data from QCEW and ?. The coefficients displayed in this table come from an event
study regression of total earnings. The dependent variable is the percent change in total earnings relative to 2001. The
variable of interest is the percent of employment that is derived from long duration industries with more than five times
more structures and intellectual property than equipment normalized to the interquartile range (IQR). Standard errors
clustered at the county level are shown in parentheses. Column (3) is shown in Figure 2d.
∗ p < 0.10, ∗∗ p < 0.05, ∗∗∗ p < 0.01.
27
Table F9: Event Study Regression of Earnings Divided by Employment on Exposure to Struc-
tures Intensive Long Duration Industries (Placebo Test)
Exposure to Placebo Industries (1) (2) (3) (4) (5)
X 1997 -0.006∗∗ -0.001 -0.001 -0.001 -0.001
(0.003) (0.002) (0.002) (0.002) (0.002)
X 1998 -0.006∗∗∗ -0.001 -0.002 -0.002 -0.001
(0.002) (0.002) (0.002) (0.002) (0.002)
X 1999 -0.004∗∗∗ -0.001 -0.000 -0.000 -0.001
(0.001) (0.001) (0.001) (0.001) (0.001)
X 2000 -0.001 0.001 0.001 0.000 -0.000
(0.001) (0.001) (0.001) (0.001) (0.001)
X 2002 0.002∗ 0.000 0.001 0.000 0.001
(0.001) (0.001) (0.001) (0.001) (0.001)
X 2003 0.005∗∗ 0.000 0.000 0.000 0.000
(0.003) (0.002) (0.001) (0.001) (0.002)
X 2004 0.012∗∗∗ 0.002 0.002 0.003 0.002
(0.004) (0.002) (0.002) (0.002) (0.003)
X 2005 0.019∗∗∗ 0.004 0.003 0.004 0.003
(0.007) (0.003) (0.003) (0.003) (0.004)
X 2006 0.024∗∗∗ 0.004 0.002 0.003 0.002
(0.008) (0.005) (0.004) (0.005) (0.005)
X 2007 0.027∗∗∗ 0.005 0.001 0.002 0.000
(0.009) (0.005) (0.005) (0.005) (0.006)
X 2008 0.026∗∗∗ 0.005 0.000 0.001 -0.001
(0.009) (0.005) (0.005) (0.006) (0.006)
X 2009 0.020∗∗∗ 0.002 -0.003 -0.003 -0.004
(0.008) (0.005) (0.005) (0.006) (0.006)
X 2010 0.019∗∗∗ 0.003 -0.004 -0.003 -0.003
(0.007) (0.005) (0.005) (0.006) (0.006)
X 2011 0.019∗∗∗ 0.002 -0.005 -0.003 -0.005
(0.007) (0.005) (0.006) (0.007) (0.007)
X 2012 0.019∗∗ 0.002 -0.005 -0.002 -0.006
(0.007) (0.006) (0.006) (0.007) (0.007)
3-digit Industry-by-Year Fixed Effects Yes Yes Yes Yes Yes
State-by-Year Fixed Effects Yes Yes Yes Yes
County Characteristics Yes Yes Yes
Winsorized Weights Yes
Drops Small County-Sectors (<1000) Yes
Note: Author’s calculations using data from QCEW and ?. The coefficients displayed in this table come from an
event study regression of total earnings divided by employment. The dependent variable is the percent change in total
earnings divided by employment relative to 2001. The variable of interest is the percent of employment that is derived
from long duration industries with more than five times more structures and intellectual property than equipment
normalized to the interquartile range (IQR). Standard errors clustered at the county level are shown in parentheses.
Column (3) is shown in Figure 2d.
∗ p < 0.10, ∗∗ p < 0.05, ∗∗∗ p < 0.01.
28
File and source
- File
- tax-policy-and-local-labor-market-behavior.pdf
- Size
- 5,036,375 bytes
- SHA-256
- 84ae86df73c8f874ab73bf70b5783d07683a8b3498fcdb68576116bc165aef61
- Original
- doi.org