REVISIONS IN UTILIZATION-ADJUSTED TFP AND ROBUST - Am MIT spezialisierte KI-Forschung

REVISIONS IN UTILIZATION-ADJUSTED TFP AND ROBUST
IDENTIFICATION OF NEWS SHOCKS

André Kurmann and Eric Sims*

Abstract—This paper documents large revisions in a widely used series of
utilization-adjusted total factor productivity (TFP) by Fernald (2014) Und
shows that these revisions can materially affect empirical results about the
effects of news shocks. We trace these revisions to changes in estimated
factor utilization that are evocative of cyclical measurement issues with
productivity. We propose an alternative identiﬁcation that is robust to these
measurement issues. Applied to U.S. Daten, the shock predicts delayed pro-
ductivity growth while simultaneously generating strong responses of novel
indicators of technological innovation and forward-looking variables. Der
shock does not lead to comovement in macroeconomic aggregates.

ICH.

Einführung

ECONOMISTS have long argued that changes in expecta-

tions about future fundamentals are an important source
of economic ﬂuctuations. This view has reemerged recently
in part due to an inﬂuential paper by Beaudry and Portier
(2006), who report that news shocks about future productiv-
ity are closely related to innovations driving long-run varia-
tions in productivity and constitute one of the main drivers
of business cycles. While the importance of news shocks for
business cycle ﬂuctuations continues to be debated, the main
identifying restriction behind news shocks is almost univer-
sally accepted: productivity reacts to news shocks only with
a delay.1

In diesem Papier, we critically revisit the zero impact restric-
tion. We argue that popular measures of productivity are
likely to be confounded by business cycle ﬂuctuations due
to imperfect measurement of factor utilization. Infolge,
news shock identiﬁcations that rely on short-run restrictions,
in particular the zero impact restriction, can produce mis-
leading results. We then propose an alternative identiﬁcation
that is robust to cyclical mismeasurement of productivity and
apply it to U.S. Daten.

The starting point of our investigation is the quarterly
utilization-adjusted series of total factor productivity (TFP)
constructed by Fernald (2014) that has become the main mea-

Received for publication November 7, 2017. Revision accepted for pub-

lication November 4, 2019. Editor: Yuriy Gorodnichenko.

∗Kurmann: Drexel University; Sims: University of Notre Dame and

NBER.

This paper combines the previous drafts by Sims (“Differences in Quar-
terly Utilization-Adjusted TFP by Vintage, with an Application to News
Shocks," Marsch 2016) and Kurmann and Otrok (“New Evidence on the
Relationship between News Shocks and the Slope of the Term Structure,”
Juni 2016). We are grateful to John Fernald for helpful conversations and
for generously sharing his code and data. We also thank Chris Otrok for ear-
lier involvement, Susanto Basu and Silvia Miranda-Agrippino for thought-
ful discussions, as well as Rudi Bachmann, Deokwoo Nam, Yuriy Gorod-
nichenko, several anonymous referees, and many seminar participants for
Kommentare.

A supplemental appendix is available online at https://doi.org/10.1162/

rest_a_00896.

1See Beaudry and Portier (2014) and Barsky, Basu, und Lee (2015) für

excellent reviews of this literature.

sure of productivity in the news literature. Fernald frequently
updates the adjusted TFP series based on new data and, weniger
frequently, implements methodological changes. We docu-
ment that a switch in detrending methods in the estimation of
utilization signiﬁcantly changes the cyclical properties of this
Serie. The sensitivity of adjusted TFP to a seemingly small
change such as this suggests, as Fernald (2014) acknowl-
edged, but otherwise mostly ignored by the literature, Das
measurement issues with productivity can be quantitatively
important.

To assess the consequences of Fernald’s revisions for news
shock identiﬁcation, we redo the estimation of Barsky and
Sims (2011), which has emerged as one of the most popular
identiﬁcation approaches in the literature. Based on prere-
vision vintages of the adjusted TFP series, a positive news
shock leads to a jump in consumption on impact but an initial
decline in hours worked. Infolge, the implied conditional
correlation of consumption growth with hours growth is neg-
ative, leading Barsky and Sims (2011) to conclude that news
shocks do not constitute a main driver of business cycles.
Based on postrevision vintages constructed with the new es-
timate of utilization, in contrast, a positive news shock leads
to a coincident increase in consumption, hours, and other real
aggregates, thereby affording a news-driven interpretation of
business cycles as Beaudry and Portier (2006) proposed.

To interpret these results and illustrate the consequences of
productivity mismeasurement for news shock identiﬁcation
more generally, we consider a medium-scale New Keyne-
sian business cycle model that allows for multiple sources
of unobserved factor utilization. Under certain conditions,
Fernald’s estimate of utilization coincidences with factor uti-
lization in the model and adjusted TFP provides an almost
perfect measure of true productivity. But under alternative yet
equally plausible conditions, Fernald’s estimate of utilization
and therefore productivity is confounded by substantial cycli-
cal mismeasurement. We conduct Monte Carlo simulations to
study the quantitative signiﬁcance of this mismeasurement.
The main insight from these simulations is that identiﬁca-
tions relying on short-run restrictions and, insbesondere, Die
zero impact restriction can be highly sensitive to differences
between factor utilization in the model and its estimation by
the econometrician. Since factor utilization is not observed
directly in the data and different assumptions about factor
utilization are difﬁcult to test, a more fruitful approach de-
vises alternative identiﬁcation restrictions that are robust to
cyclical mismeasurement.

In the ﬁnal section of the paper, we propose such an alterna-
tive identiﬁcation. Building on the premise by Beaudry and
Portier (2006) that news shocks capture information about
slowly disseminating changes in technology and economic
organization that drive long-run productivity, we extract the

The Review of Economics and Statistics, Mai 2021, 103(2): 216–235
© 2020 by the President and Fellows of Harvard College and the Massachusetts Institute of Technology
https://doi.org/10.1162/rest_a_00896

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
R
e
S
T
/

A
R
T
ich
C
e
–
P
D

F
/

1
0
3
2
2
1
6
1
9
1
5
8
9
8
/
R
e
S
T
_
A
_
0
0
8
9
6
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

ROBUST IDENTIFICATION OF NEWS SHOCKS

217

innovation that accounts for the maximum forecast error vari-
ance (FEV) share of adjusted TFP at a long but ﬁnite horizon.2
This max-share approach, which builds on work by Uhlig
(2003), has been used previously by Francis et al. (2014) Zu
identify long-run technology shocks. We differ in that we ap-
ply it to adjusted TFP instead of labor productivity and pro-
pose it as a possible news identiﬁcation. Conceptually, Die
max-share identiﬁcation is also similar to Barsky and Sims
(2011) and many close variants in the news literature, mit dem
crucial difference, Jedoch, that it does not impose the zero
impact restriction and, more generally, does not rely on short-
run ﬂuctuations in productivity. The max-share identiﬁcation
should therefore be more robust to cyclical mismeasurement
of productivity, and we verify this through Monte Carlo sim-
ulations with our model.3 In these simulations, the max-share
identiﬁcation performs very well as long as mean-reverting
surprise technology shocks do not account for a large fraction
of the lower-frequency variation in true technology.

Natürlich, nothing guarantees that the max-share identi-
ﬁcation captures news shocks as opposed to other shocks
driving future productivity. Jedoch, when applied to U.S.
Daten, we ﬁnd compelling evidence in favor of a news interpre-
Station. The shock has no signiﬁcant impact on adjusted TFP
for several quarters but predicts sustained future productivity
Wachstum, accounting for 70% or more of TFP ﬂuctuations at
long forecast horizons. More important, the shock is associ-
ated with large impact responses of two novel indicators of
Innovation: an index of books published in the ﬁelds of tech-
nology by Alexopoulos (2011) and an index of technological
standardization by Baron and Schmidt (2015), followed by
a hump-shaped increase in R&D expenditures and a grad-
ual decline in the relative price of investment goods. Dritte,
the shock generates strong, positive immediate reactions of
forward-looking information variables.

In terms of macroeconomic implications, the max-share
identiﬁcation implies very similar impulse responses as the
ones that Barsky and Sims (2011) originally reported, mit
the important difference that all the results are robust to the
revisions in Fernald’s adjusted TFP series. Consumption in-
creases on impact of the shock and then gradually rises to
a new permanent level, while hours worked initially decline
and later increase in a hump-shaped pattern before returning
to the preshock level. The shock therefore implies a negative
correlation between consumption growth and hours worked,
which makes it an unlikely source of business cycle ﬂuctu-
ations. Trotzdem, the shock accounts for a large share of
macroeconomic ﬂuctuations at medium and longer horizons
and generates sharp impact responses of inﬂation and asset
Preise.

The fact that the empirical ﬁndings from the max-share
identiﬁcation do not substantively differ from those in Barsky

2The idea that new technologies diffuse slowly ﬁnds ample support in
a large micro-empirical literature. See Griliches (1957), Mansﬁeld (1961,
1989), Gort and Klepper (1982), and Rogers (1995).

3The max-share identiﬁcation is also robust to situations in which inno-
vations to expected future productivity have an immediate impact on (true)
productivity.

and Sims (2011) does not mean that mismeasurement of pro-
ductivity is without consequence. Insbesondere, es gibt kein
free lunch in the sense that because of these measurement
issues, one cannot separately identify surprise shocks to cur-
rent productivity from news shocks about future productivity.
This is important for other research that relies on short-run
ﬂuctuations in adjusted TFP for identiﬁcation.

The main lesson of the paper is that cyclical measure-
ment issues can materially affect the identiﬁcation of news
shocks based on short-run restrictions. While measurement
error occupies a central role in many ﬁelds of economics,
it has generally taken a back seat in quantitative macroeco-
nomics. A notable exception is Christiano, Eichenbaum, Und
Vigfusson (2004), who argue, as we do, that adjusted TFP
may be confounded by measurement error.4 They then ap-
ply the inﬁnite-horizon strategy of Gali (1999) to identify
long-run productivity shocks based on the assumption that
measurement errors in adjusted TFP are transient. Our paper
differs in important aspects from Christiano et al. (2004) Und
the related literature on long-run productivity shocks. Erste,
the max-share identiﬁcation proposed here does not impose
that technology is the only source of long-run ﬂuctuations in
productivity and instead extracts the shock that accounts for
the maximum FEV share of adjusted TFP at a long but ﬁnite
horizon. The max-share approach therefore affords the pos-
sibility that other shocks (z.B., a surprise productivity shock)
exert long-lasting effects on adjusted TFP and at the same
time addresses the criticism that inﬁnite-horizon restrictions
imply potentially large biases in ﬁnite-order VARs.5

Zweite, the literature on long-run productivity shocks typ-
ically uses average labor productivity as the technology mea-
sure and is primarily concerned with the dynamics of hours
worked in response to a shock.6 As such, this literature does
not directly relate to the news literature and the idea that
improvements in technology disseminate slowly and in a
predictable manner. Indeed in many cases—including the
max-share implementation by Francis et al. (2014) on which
our identiﬁcation is based—labor productivity jumps imme-
diately because hours fall on impact, thus resulting in capi-
tal deepening.7 This may lead to the inadvertent conclusion,
often imposed in DSGE models, that technology follows a
random walk process with only one shock. While our results
are consistent with the fact that empirically measured TFP
is well characterized by a univariate random walk process,

4Chang and Li (2018) is another more general investigation about the
sensitivity of recent research results to measurement error in gross domestic
product.

5Bias reduction in ﬁnite-order VARs is the main motivation of Francis et al.
(2014) for the max-share identiﬁcation. Also see Ereeg, Guerrieri, and Gust
(2005), Christiano et al. (2006), and Chari, Kehoe, and McGrattan (2008)
for important contributions in this respect. Another practical advantage of
the max-share approach is that it can be implemented with either a VAR in
levels that includes nonstationary variables, as we do, or a stationary VAR.
6Aside from Christiano et al. (2004), one other exception is Chen and
Wemy (2015) WHO, like us, apply the max-share approach to adjusted TFP.
Jedoch, they do not investigate the robustness of the approach to revisions
in adjusted TFP or whether the resulting shock is a news shock.

7We conﬁrm this result in our VAR speciﬁcation. See the discussion in

section V for details.

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
R
e
S
T
/

A
R
T
ich
C
e
–
P
D

F
/

1
0
3
2
2
1
6
1
9
1
5
8
9
8
/
R
e
S
T
_
A
_
0
0
8
9
6
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

218

THE REVIEW OF ECONOMICS AND STATISTICS

they nevertheless suggest that the permanent component of
the series is slow diffusion, which is rather different from the
oft-assumed jump process impulse response functions gen-
erated from a univariate random walk process.

Within the extensive VAR literature on news shocks, unser
paper is perhaps most closely related to the one by Barsky
et al. (2015). They identify a news shock by imposing a
longer-run restriction that is conceptually similar to the max-
share approach proposed here but differs in potentially im-
portant details.8 Using a prerevision vintage of Fernald’s ad-
justed TFP series, they ﬁnd that their news shock looks quite
similar independent of whether they impose the zero impact
restriction. Our results conﬁrm their ﬁnding in the sense that
the initial response of adjusted TFP to the proposed max-
share shock is small and insigniﬁcantly different from 0 für
several quarters. Our contribution relative to the paper by
Barsky et al. (2015) and the rest of the news literature is to
document the large revisions in Fernald’s utilization-adjusted
TFP series and show that these revisions can materially af-
fect empirical conclusions about the effects of news shocks
based on short-run restrictions.9 We propose the max-share
approach as an alternative identiﬁcation of news shocks and
show that it is robust to measurement issues, and we go to con-
siderable length to establish the news content of the extracted
shock by relating it to measures of technological innovation
and forward-looking information variables.

The idea that the slow dissemination of technology implies
predictable long-run changes in productivity relates to a re-
cent (non-news) literature on the macroeconomic effects of
persistent productivity growth processes. Rotemberg (2003)
discusses extensively the available evidence on the slow dis-
semination of technology and proposes a model in which
random technological progress leads to stochastic variations
in long-run output while deviations of output from trend are
mostly driven by temporary shocks. As in our empirical in-
vestigation, he ﬁnds that slowly diffusing technical progress
leads to a temporary drop in hours worked and economic
activity.10 Lindé (2009) incorporates autocorrelated shocks
to the growth rate of productivity into an otherwise standard

8The identiﬁcation of Barsky et al. (2015) extracts the shock that accounts
for all of the forecast revision of adjusted TFP at some long but ﬁnite horizon
subject to the zero impact restriction, although their results would be very
similar if this restriction was not imposed. We prefer the max-share approach
Weil, wie oben beschrieben, it does not impose that news shocks are the only
source of predictable ﬂuctuations at that particular horizon. Außerdem,
our Monte Carlo simulations reveal that imposing the zero impact restriction
can have important consequences even if without this restriction, the impact
response of adjusted TFP is close to 0.

9In contemporaneous work, Cascaldi-Garcia (2017) also points out that
revisions in Fernald’s adjusted TFP series affect the macroeconomic impli-
cations of news shocks based on the Barsky and Sims (2011) identiﬁcation.
The paper does not document the source of these revisions in detail or dis-
cuss why these revisions raise questions about the zero impact restriction
imposed by the news literature. Stattdessen, the paper is intended as a com-
ment on Kurmann and Otrok (2013), to which Kurmann and Otrok (2017B)
respond using the alternative identiﬁcation approach proposed here.

10Other papers that document the slow diffusion of technology and build
models of costly adoption are Comin and Gertler (2006), Comin and Hobijn
(2010), and Comin, Gertler, and Santacreu (2009).

RBC model. Autocorrelated shocks to productivity growth
have the ﬂavor of news, and he shows that incorporating this
feature can help reconcile the RBC model with empirical
results on the effects of technology shocks on hours worked.

II. Revisions in Utilization-Adjusted TFP

The business cycle literature has typically measured pro-
ductivity as the residual of aggregate output not accounted
for by capital and labor inputs, commonly known as TFP.
Economists quickly realized, Jedoch, that TFP may be a
poor proxy of technology for a variety of reasons, most no-
tably changes in unobserved factor utilization. In Beantwortung
to these concerns, Basu, Fernald, and Kimball (2006) con-
struct an aggregate measure of productivity that takes into
account sectoral heterogeneity, imperfect competition, com-
positional changes in the quality of labor and capital, and un-
observed factor utilization. Fernald (2014) extends the anal-
ysis of Basu et al. (2006), which is carried out with annual
Daten, to construct a quarterly measure of TFP. Because of
the higher frequency, not all of the corrections in the original
Basu et al. (2006) series can be implemented, but perhaps
the most important one—the adjustment for variable factor
utilization—is.

In what follows, we brieﬂy review the construction of
Fernald’s utilization-adjusted TFP series. We then document
how seemingly small changes in the estimation of factor uti-
lization lead to large revisions in utilization-adjusted TFP that
materially affect its business cycle properties.

A. Fernald’s Utilization-Adjusted TFP Series

Fernald’s series of utilization-adjusted TFP is based on the
assumption that there exists an aggregate production function

Yt = F (et Lt , zt Kt , Bei ),

(1)

whereYt denotes output, Lt labor input (the product of average
hours per worker, ht , and employment, Nt , adjusted for quality
of the workforce),11 Kt capital input, et labor effort, zt capital
use, and At technology that should be understood broadly as
a shifter of the production function.

Differentiating equation (1) with respect to time and fur-
ther assuming constant returns to scale as well as price taking
by ﬁrms in perfectly competitive input and output markets,
cost minimization implies that technology growth can be ex-
pressed as

˙At
Bei

(cid:2)

˙Yt
Yt

− ωL,T

˙Lt
Lt

− ωK,T

(cid:3)

(cid:2)
ωL,T

−

˙et
et

˙Kt
Kt

+ ωK,T

(cid:3)

˙zt
zt

(2)

where ωL,t denotes the cost share of labor and ωK,t the cost
share of capital, which under constant returns to scale equals

11Fernald adjusts for changes in workforce quality using estimates of

education and experience for different groups.

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
R
e
S
T
/

A
R
T
ich
C
e
–
P
D

F
/

1
0
3
2
2
1
6
1
9
1
5
8
9
8
/
R
e
S
T
_
A
_
0
0
8
9
6
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

ROBUST IDENTIFICATION OF NEWS SHOCKS

219

TABLE 1.—MOMENTS OF TFP GROWTH AND UTILIZATION FOR DIFFERENT VINTAGES

A. Adjusted TFP Growth

(cid:2) ln TFPu,07

(cid:2) ln TFPu,13

(cid:2) ln TFPu,14

(cid:2) ln TFPu,16

Mean
Standard deviation
Correlation with (cid:2) ln TFPu,07
Correlation with (cid:2) ln Yt
Correlation with (cid:2) ln Lt

1.49
3.41
1.00
0.53
−0.01

1.41
3.30
0.85
0.38
−0.06

1.42
3.79
0.56
0.18
−0.24

1.42
3.46
0.58
0.07
−0.35

Mean
Standard deviation
Correlation with (cid:2) ln TFPu,07

Mean
Standard deviation
Correlation with (cid:2) ln (cid:4)u07
T
(cid:2) ln TFPu, J

(cid:2) ln TFP07
T

(cid:2) ln TFP13
T

(cid:2) ln TFP14
T

(cid:2) ln TFP16
T

B. Unadjusted TFP Growth

1.42
3.75
1.00

(cid:2) ln (cid:4)u07
T
−0.08
2.34
1.00

1.37
3.55
0.92

(cid:2) ln (cid:4)u13
T
−0.04
2.94
0.94

C. Utilization Growth

1.37
3.55
0.92

(cid:2) ln (cid:4)u14
T
−0.05
3.75
0.58

1.39
3.55
0.93

(cid:2) ln (cid:4)u16
T
−0.03
3.76
0.65

is the quarterly log change expressed in annualized percentage points of Fernald’s adjusted TFP series for vintages j = 07, 13, 14, oder 16; (cid:2) ln TFP j

t is unadjusted TFP growth by vintage; Und (cid:2) ln (cid:4)u j
T
is the growth rate of Fernald’s utilization series by vintage. Yt is real GDP and Ht is total hours worked in the nonfarm business sector; these are from the NIPA tables and are expressed as quarterly log changes. Der
sample period for each of the statistics is 1947q3 to 2007q3.

(1 − ωL,T ). The term in the ﬁrst parenthesis is typically re-
ferred to as TFP growth and the term in the second parenthesis
as the change in factor utilization.

Fernald constructs TFP growth from quarterly NIPA and

BLS data as

(cid:2) ln TFPt = (cid:2) ln Yt − ωL,T (cid:2) ln Lt − (1 − ωL,T )(cid:2) ln Kt ,

(3)

with output growth measured as the log change in the equally
weighted average of real expenditures and income in the
business sector; and labor and capital growth built up from
quality-adjusted series of different labor and capital types.
To adjust for variable labor effort and capital use, welche
are not directly observed in the data, Fernald follows Basu
et al. (2006) and proxies the change in factor utilization by a
weighted change in industry hours per worker, das ist,

Given equations (3) Und (4), utilization-adjusted TFP,

(cid:2) ln TFPu
T

= (cid:2) ln TFPt − (cid:2) ln ˆut ,

(5)

provides an empirical estimate of aggregate technology
growth as deﬁned in equation (1). As Fernald (2014) explic-
itly acknowledged, with markups, possibly heterogeneous
across producers, of price above marginal cost, or with factor
adjustment costs that lead the shadow cost of inputs to dif-
fer across ﬁrms . . . aggregate TFP and aggregate technology
are not the same—even in the absence of variable factor uti-
lization.” Similarly, if the utilization proxy in equation (4) Ist
incorrect, then this will also lead to mismeasurement. Despite
these potential issues, adjusted TFP is an important bench-
mark and has become the primary measure of technology for
business cycle macroeconomics.

Δ ln ˆut =

(cid:5)

ich

κi

(cid:4)βiΔ ln hc

(4)

B. Changes across Vintages

where hc
it denotes a measure of hours per worker in indus-
try i discussed further below; κi the industry weights; Und
(cid:4)βi the industry-speciﬁc factors of proportionality estimated
using demand-side shocks as instruments.12 The idea behind
this proxy is that industry capital stocks and employment are
quasi-ﬁxed, but hours per worker, labor effort, and capital
use can be adjusted costlessly. Under certain conditions, Re-
viewed in detail in section IV, optimal ﬁrm behavior then
implies that utilization is proportional to hours per worker.

12See Fernald (2014) for details on the data and instrumental variable

estimation procedure.

Fernald regularly publishes revised estimates of adjusted
TFP based on new data and methodological changes.13 Panel
a of table 1 reports key statistics for the vintages of December
2007, Dezember 2013, Mai 2014, and May 2016, all over the
same sample period 1947:3 Zu 2007:3.14 The means and stan-
dard deviations are similar across vintages. Jedoch, Dort
is a marked change in business cycle comovement from the
2014 vintage onward. The correlation coefﬁcient between the

13The different vintages of adjusted TFP, as well as the underlying com-

ponents, are available on Fernald’s website.

14The beginning and end of the sample are dictated by the availability of
the December 2007 vintage. The results for other pre-2014 vintages are very
similar to the 2007 Und 2013 vintages, while the results for other post-2013
vintages are very similar to the 2014 Und 2016 vintages.

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
R
e
S
T
/

A
R
T
ich
C
e
–
P
D

F
/

1
0
3
2
2
1
6
1
9
1
5
8
9
8
/
R
e
S
T
_
A
_
0
0
8
9
6
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

220

THE REVIEW OF ECONOMICS AND STATISTICS

Standard deviation
Correlation with (cid:2) ln (cid:4)u07
T

TABLE 2.—CHANGES IN FERNALD’S UTILIZATION ESTIMATES

(cid:2) ln (cid:4)u13
T

2.94
0.94

(cid:2) ln (cid:4)u13,BF F K
T

2.26
0.91

(cid:2) ln (cid:4)u13,BW
T

4.76
0.59

(cid:2) ln (cid:4)u13,BF F K&BW

3.69
0.57

(cid:2) ln (cid:4)u14
T

3.75
0.58

This table shows simulated utilization series based on the 2013 vintage data. See the text for details. The sample period for all statistics is 1947q3 to 2007q3.

2007 vintage and post-2013 vintages of adjusted TFP growth
is less than 0.6. Außerdem, während die 2007 Und 2013 vin-
tages of adjusted TFP growth are positively correlated with
output growth and uncorrelated with total hours growth, Die
2014 Und 2016 vintages of adjusted TFP growth are uncorre-
lated with output growth and negatively correlated with total
hours growth.15

Panels b and c in table 1 decompose these changes in corre-
lation into the parts coming from unadjusted TFP and utiliza-
tion. The business cycle properties of unadjusted TFP growth
remain essentially unchanged across vintages. Variations in
utilization, by contrast, are signiﬁcantly larger for the 2014
Und 2016 vintages, and there is an important decline in cor-
relation relative to the 2007 Und 2013 vintages.16

This suggests that the large changes in business cycle prop-
erties of adjusted TFP are not due to revisions in nonadjusted
TFP (d.h., due to data revisions or changes in NIPA methodol-
Ogy), but are instead driven primarily by revisions in utiliza-
tion. We conﬁrm this conjecture by combining the 2007 vin-
tage of utilization with nonadjusted TFP from other vintages.
Correlations for the resulting synthetic series of adjusted TFP
are presented in the online appendix. The correlations of the
2007 adjusted TFP vintage with the synthetic 2014 Und 2016
series are both 0.91, compared to 0.56 Und 0.58 for the actual
vintages. Somit, while revisions in utilization do not account
for all of the changes in adjusted TFP, they explain the large
majority.

C. Revisiting Fernald’s Estimation of Utilization

What explains the changes in estimated utilization across
vintages? Between December 2013 and May 2014, Fernald
implemented two methodological changes. Erste, he switched
from using estimates of industry weights and proportionality
factors (cid:4)βi in equation (4) by Basu et al. (2006) to estimates
from Basu et al. (2013), which are based on more recent data
and a more detailed industry decomposition. Zweite, Fernald
has to contend with the issue that hours per worker in many
industries are trending over time. Up to the December 2013
vintage, Fernald follows Basu et al. (2006) and detrends in-
dustry hours per worker with the bandpass ﬁlter of Christiano

15These changes in correlation across vintages of adjusted TFP occur for

different subsamples and are not driven by a particular time period.

16Visual inspection of the series for unadjusted and adjusted TFP along
with estimated factor utilization (see the appendix) suggests that changes
in nonadjusted TFP across vintages cannot account for these large changes
in the adjusted TFP series. Eher, there seem to be noticeable changes in
estimated utilization across vintages, mit dem 2007 vintage substantially
smoother than less persistent than the 2016 vintage.

and Fitzgerald (2003) to isolate frequencies between 8 Und 32
quarters. From the May 2014 vintage onward, Fernald instead
detrends industry hours per worker with the bi-weight ﬁlter
used in Stock and Watson (2012), which removes a much
slower-moving trend than the bandpass ﬁlter.

Using replication codes for the December 2013 and May
2014 vintages shared generously by Fernald, we assess the
quantitative importance of the two changes. Tisch 2 Berichte
the results. For comparison, the ﬁrst and the last columns
replicate the business cycle properties of the actual December
2013 and May 2014 vintages of estimated utilization growth
from table 1. The second column, beschriftet (cid:2) ln ˆu13,BF F K
,
shows the effect of switching to the industry weights and
proportionality factors from Basu et al. (2013). While this
switch lowers the volatility of utilization somewhat, it leaves
the correlation with the 2007 vintage essentially unchanged.
As shown by the third column, beschriftet (cid:2) ln ˆu13,BW
, by con-
trast, changing the detrending method from bandpass ﬁlter-
ing to bi-weight ﬁltering leads to a substantial increase in the
volatility of utilization growth and a concurrent decrease in
the correlation with the 2007 vintage. Endlich, as shown in the
fourth column, beschriftet (cid:2) ln ˆu13,BF F K&BW
, the two changes es-
T
sentially replicate the 2014 vintage. The remaining difference
is due to data revisions.

The results make clear that the change in ﬁltering of hours
per worker is the main driver of the revisions in utilization
Wachstum. We wish to emphasize, Jedoch, that it is not the
bandpass or bi-weight ﬁlter per se that matters for the results,
but rather the frequencies isolated by the different ﬁlters. Als
notiert, the bandpass ﬁlter used in the earlier vintages of Fer-
nald’s series removes high-frequency ﬂuctuations from uti-
lization, whereas the bi-weight ﬁlter does not. It is the inclu-
sion of the higher-frequency ﬂuctuations with the bi-weight
ﬁlter, not the relatively lower-frequency ﬂuctuations that it
captures compared to the bandpass ﬁlter, that account for the
differences across vintages.17

For the purpose of the news identiﬁcation that follows, Es
is not clear that either the bandpass or the bi-weight ﬁlter,
or any other statistical detrending method for that matter,
adequately captures the appropriate ﬂuctuations in hours per
worker to correctly infer factor utilization. This means that for
either ﬁltering choice, utilization and therefore adjusted TFP
may still be confounded by cyclical mismeasurement even
if the conditions underlying the proportionality assumption

17In der Tat, experimentation with alternative values of the ﬁltering parame-
ters conﬁrms that the important difference is that the bandpass ﬁlter removes
higher-frequency ﬂuctuations, whereas the bi-weight ﬁlter leaves them in.

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
R
e
S
T
/

A
R
T
ich
C
e
–
P
D

F
/

1
0
3
2
2
1
6
1
9
1
5
8
9
8
/
R
e
S
T
_
A
_
0
0
8
9
6
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

ROBUST IDENTIFICATION OF NEWS SHOCKS

221

in equation (4) are satisﬁed (a point to which we return in
section IV).

III.

Implications for News Shocks Identiﬁcation

Starting with Cochrane (1994),

the modern macro-
literature has deﬁned news shocks as information useful in
predicting future fundamentals (often productivity) but un-
related to current and past fundamentals. As proposed by
Beaudry and Portier (2006) proposed, this implies a zero im-
pact restriction, which is that news shocks affect productivity
only with a delay. This restriction is at the core of almost all
news shocks identiﬁcations used to date.

In what follows, we use the news identiﬁcation approach
of Barsky and Sims (2011) to quantify the implications of
the revisions in adjusted TFP for news shocks, although
the lessons learned are relevant for other identiﬁcation ap-
proaches relying on the zero impact restriction as well. Wir
focus on the Barsky-Sims approach because it does not re-
quire taking a stand on the nature of non-news shocks.18 Fur-
thermore, Monte Carlo simulations show that the Barsky-
Sims approach performs well provided that productivity is
measured correctly. Als solche, the Barsky-Sims approach has
emerged as one of the most commonly used identiﬁcations
in the literature.

A. Barsky-Sims Identiﬁcation

The Barsky-Sims identiﬁcation consists of estimating a
VAR and extracting the innovation that is orthogonal to Fer-
nald’s adjusted TFP series but maximally accounts for the
FEV share of adjusted TFP over a ten-year horizon. Seit
our alternative identiﬁcation proposed in section V is con-
ceptually very similar, we review the details here. Let Yt be
a k × 1 random vector process of which the ﬁrst variable is a
measure of productivity (z.B., Fernald’s utilization-adjusted
TFP), and let the reduced-form moving average representa-
tion of this process be given by Yt = B(L)ut , where ut is
a k × 1 vector of prediction errors with variance-covariance
matrix E (ut u(cid:3)
T ) = (cid:2)u, and B(L) = I + B1L + B2L2 + . . . Ist
a matrix lag polynomial.

Now assume that there exists a linear mapping between the
prediction errors and the structural shocks, ut = A(cid:4)T , Wo
(cid:4)t is a k × 1 vector of structural shocks characterized by
E ((cid:4)T (cid:4)(cid:3)
T ) = I, and A is a k × k matrix satisfying AA(cid:3) = (cid:2)u.
Given the symmetry of (cid:2)u, a multitude of A is consistent with
AA(cid:3) = (cid:2)u. The Choleski decomposition of (cid:2)u is one poten-

tial solution. Denote this by (cid:6)A. The entire set of permissible
values of A consistent with AA(cid:3) = (cid:2)u is then described by
(cid:6)AQ, where Q is an orthonormal rotation matrix and the struc-
tural moving average representation is Yt = C(L)(cid:4)T , Wo
C(L) = B(L)(cid:6)AQ.

The h step ahead forecast error of Yt can be written as

Yt+h − Et−1Yt+h =

H(cid:5)

l=0

(cid:6)AQ(cid:4)t+h−l .

(6)

The FEV share of variable i attributable to shock j at horizon
h is then

(cid:3)ich, J (H) =

(cid:7)

H
l=0 Bi,l
(cid:7)
H

(cid:6)Aγγ(cid:3)(cid:6)A(cid:3)B(cid:3)
ich,l

l=0 Bi,l (cid:2)uB(cid:3)
ich,l

(7)

where Bi,l is the ith row of lag polynomial evaluated at L = l
and γ is the jth column of Q.

The news shock identiﬁcation of Barsky and Sims (2011)
consists of picking γ to maximize the sum of FEV shares
of productivity (the ﬁrst variable in the VAR) up to some
truncation horizon H subject to the restriction that the shock
is orthogonal to current productivity.

Formally,

max
γ

H(cid:5)

h=0

(cid:5)1,2(H) s.t. γ(cid:3)γ = 1 and γ(1, 1) = 0,

(8)

where without loss of generality, productivity is ordered ﬁrst
in Yt and the news shock is deﬁned as the second shock in (cid:4)T .
The ﬁrst restriction ensures that γ belongs to an orthonormal
Matrix. The second restriction imposes that the news shock
affects productivity only with a delay.

B. Effect of Revisions on News Shock Identiﬁcation

We apply the Barsky-Sims identiﬁcation to a four-variable
VAR comprising either the 2007 vintage or the 2016 vintage
of Fernald’s utilization-adjusted TFP series, real personal
consumption expenditures per capita, total hours worked per
capita in the nonfarm business sector, and inﬂation as mea-
sured by the growth rate of the GDP price deﬂator.19 Re-
sults for larger VARs that contain additional macro aggre-
gates are similar. With the exception of the inﬂation rate, Die
variables enter the VAR in log levels. The VAR is estimated
with four lags via Bayesian methods subject to a Minnesota
prior.20 Conﬁdence bands are computed by drawing from the

18The zero impact restriction is sufﬁcient to identify news shocks in bi-
variate VARs. In VARs with more than two variables, additional restrictions
need to be imposed. Full identiﬁcation approaches that do so by taking a
stand on all structural shocks affecting the VAR are often subject to im-
portant robustness issues. Sehen, Zum Beispiel, Kurmann and Mertens (2014),
who show that the identiﬁcation by Beaudry and Portier (2006) nicht
have a unique solution in their VAR systems with more than two variables,
or Fisher (2010), who shows that the results of Beaudry and Lucke (2010)
depend on the choice of cointegration restrictions imposed.

19VARs based on any of the pre-2014 vintages of adjusted TFP produce
impulse responses that are nearly identical to those based on the 2007 vin-
tage, while VARs based on post-2014 vintages of adjusted TFP produce
impulse responses that are very similar to those based on the 2016 vintage.
20The Minnesota prior assumes a random walk process for adjusted TFP
and consumption and a white noise process for hours worked and the in-
ﬂation rate. Estimates are robust to assuming a random walk prior for all
Serie.

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
R
e
S
T
/

A
R
T
ich
C
e
–
P
D

F
/

1
0
3
2
2
1
6
1
9
1
5
8
9
8
/
R
e
S
T
_
A
_
0
0
8
9
6
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

222

THE REVIEW OF ECONOMICS AND STATISTICS

resulting posterior distribution. The sample period is ﬁxed
at 1960q1 to 2007q3.21 As in Barsky and Sims (2011), Die
truncation horizon is set to H = 40.

Figure 1a presents impulse responses to a news shock us-
ing the Barsky and Sims (2011) news identiﬁcation. Hier
and below, the solid lines show the posterior median impulse
responses implied by the posterior distribution of the VAR es-
timated with the 2016 vintage of adjusted TFP, and the gray
bands are the corresponding 16% Zu 84% posterior coverage
Intervalle. Im Gegenzug, the dash-dotted lines show the posterior
median impulse responses implied by the posterior distribu-
tion of the VAR estimated with the 2007 vintage of adjusted
TFP, and the dashed lines are the corresponding 16% Zu 84%
posterior coverage intervals.

Based on the 2007 vintage of adjusted TFP, the responses
are very similar to those estimated by Barsky and Sims
(2011). Adjusted TFP starts to increase the quarter after
the shock, consumption jumps up while inﬂation falls on
impact, and hours worked initially decline, turning signiﬁ-
cantly positive only after about twelve quarters. As shown
in the appendix, these responses imply that if the economy
was buffeted solely by news shocks, the correlation between
consumption growth and hours growth would be negative,
whereas in the data, this correlation is robustly positive.

Based on the 2016 vintage of adjusted TFP, in contrast, Die
impulse responses look different in economically meaning-
ful ways. Adjusted TFP reacts to the news shock only after
several quarters, while hours worked increase from the be-
ginning (although insigniﬁcantly for the ﬁrst few quarters),
reaching peak response about ten quarters earlier than based
on the 2007 vintage. This difference in the response of hours
worked implies that the correlation of consumption growth
and hours growth conditional on news shocks is now positive
(see the appendix for details). Darüber hinaus, the deﬂationary im-
pact of news shocks, which Barsky et al. (2015) cite as one of
the most robust features of the data, is no longer statistically
signiﬁcant.

The difference in responses depending on the vintage of
adjusted TFP used has important implications for the role
of news shocks. Based on the 2007 vintage, the absence of
comovement between hours and consumption leads Barsky
and Sims (2011) to conclude that news shocks about future
productivity are not a major source of business cycle ﬂuctu-
ations. Based on the 2016 vintage instead, the coincident in-
crease in consumption and hours is consistent with the view
espoused by Beaudry and Portier (2006) that news shocks
have signiﬁcant short-term demand effects and are a poten-
tially important driver of business cycle ﬂuctuations.

21The beginning of the sample is chosen to facilitate comparison with
Barsky and Sims (2011) and because some additional variables of interest
studied below are unavailable prior to 1960. Außerdem, the omission of
the immediate postwar data from the sample removes some large inﬂuences
due the 1951 Treasury Accord and Korean War. The end date is the last
available observation for the 2007 vintage of adjusted TFP data.

IV.

Interpreting the Results through a DSGE Model

The results of the preceding sections illustrate that mea-
surement issues about productivity can have important impli-
cations for news shock identiﬁcations that rely on short-run
restrictions on productivity and, insbesondere, on the zero im-
pact restriction. To interpret and better understand these re-
Ergebnisse, we build a medium-scale New Keynesian DSGE model
and conduct different Monte Carlo simulations to address the
following questions. Under what conditions does Fernald’s
adjusted TFP series appropriately measure technology? Was
are the consequences of different sources of productivity mis-
measurement for news shock identiﬁcation? Is Fernald’s new
bi-weight ﬁltered estimate of utilization preferable to the pre-
viously used bandpass ﬁltered estimate?

The DSGE model we use is based on Christiano, Eichen-
baum, and Evans (2005), Smets and Wouters (2007), Und
Justiniano, Primiceri, and Tambalotti (2010) but augmented
with variable labor effort and hours worked so as to analyze
the conditions under which the proportionality result between
utilization and hours worked set forth in Basu et al. (2006)
holds. The model abstracts from heterogeneity in production
across industries and imposes an aggregate production func-
tion. While potentially important, these assumptions do not
invalidate the measurement issues highlighted here.

A. Modell

To save on space, we focus on the model components that
relate to the measurement of technology and utilization. A
full description of the model is provided in the appendix.

The model is populated by intermediate goods producers,
a representative ﬁnal goods producer, a representative house-
hold, labor unions, a labor packer, and a monetary authority.
Intermediate goods producers are indexed by i ∈ [0 1] Und
produce output with

Yt (ich) = At

(cid:8)
Ks,T (ich)

(cid:9)α (cid:8)

(cid:9)
Ls,T (ich)

1−α − F Xt ,

(9)

where At denotes exogenous technology (common across
ﬁrms), Ks,T (ich) capital services, Ls,T (ich) labor services, Und
F Xt ≥ 0 is a ﬁxed cost that increases with the economy’s
trend Xt . Intermediate outputs are aggregated into ﬁnal out-
put via a CES technology, and intermediate producers are
subject to the typical Calvo pricing friction.

Log technology is the sum of two components, ln At =

ln St + ln (cid:6)T , where St follows

ln St = ρS ln St−1 + σSεS,T ,

(10)

with εS,t i.i.d. (0,1), Und (cid:6)t is a permanent component that
evolves according to

ln (cid:6)t − ln (cid:6)t−1 = (1 − ρ(cid:6) ) ln g + ρ(cid:6) (ln (cid:6)t−1 − ln (cid:6)t−2)

+ σgεg,t−1,

(11)

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
R
e
S
T
/

A
R
T
ich
C
e
–
P
D

F
/

1
0
3
2
2
1
6
1
9
1
5
8
9
8
/
R
e
S
T
_
A
_
0
0
8
9
6
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

ROBUST IDENTIFICATION OF NEWS SHOCKS

223

FIGURE 1.—IMPULSE RESPONSES TO NEWS SHOCK, FOUR VARIABLE VAR, 2007 VERSUS 2016 VINTAGE

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
R
e
S
T
/

A
R
T
ich
C
e
–
P
D

F
/

1
0
3
2
2
1
6
1
9
1
5
8
9
8
/
R
e
S
T
_
A
_
0
0
8
9
6
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

News shocks for different vintages of TFP via (A) Barsky-Sims and (B) the max-share. Solid lines are the posterior median estimates from the VAR system estimated with the 2016 vintage of adjusted TFP. The shaded
bands correspond to the 16% Zu 84% posterior coverage intervals. The dash-dotted lines are the posterior median estimates for the system estimated with the 2007 vintage of adjusted TFP; the dashed lines correspond
to the 16% Zu 84% posterior coverage intervals. The sample period is 1960q1 to 2007q3.

224

THE REVIEW OF ECONOMICS AND STATISTICS

with εg,t−1 i.i.d. (0,1). This shock is assumed to occur before
it starts to have an impact on technology and agents update
expectations about the permanent component accordingly.
Darüber hinaus, since ρ(cid:6) > 0, this shock portends even larger in-
creases in the level of technology in the future.

The representative household consists of a continuum of
members, a fraction Nt of whom are working and a fraction
1 − Nt who are not working. Employed members provide
labor services Lt = et ht Nt to labor unions, where ht denotes
average hours worked and et is labor effort. Members of the
household are randomly chosen to work, with the household
head choosing the total fraction of workers, Nt . All workers
supply the same hours and effort, and all members enjoy the
same consumption regardless of whether they work or not
(d.h., there is perfect intrahousehold insurance). The expected
lifetime utility of the household is

(cid:10)

βt νt

∞(cid:5)

t=0

ln (Ct − bCt−1) + θNt (T − G(ht , et ))

+ θ(1 − Nt )T

(cid:11)
,

at nominal wage rate W l
nominal wage rigidity.

T . Unions are subject to Calvo-style

Endlich, the monetary authority sets the nominal interest
rate according to a conventional inertial Taylor-type rule in
inﬂation and output growth.

To assess the conditions under which Fernald’s adjusted
TFP series accurately measures technology, we start with
unadjusted TFP as deﬁned in equation (3). Even if utiliza-
tion was constant (or variations in utilization were perfectly
corrected), the model would nevertheless imply two incon-
gruities between adjusted TFP and technology. Erste, mit
ﬁxed cost F > 0, the production function is not constant re-
turns to scale, thus invalidating the assumption that the cost
shares of labor and capital sum up to 1 (d.h., ωL,T + ωK,t = 1).
Zweite, if intermediate goods ﬁrms have market power and
are subject to nominal price rigidities, then ωL,t and 1 − ωL,T
do not in general correspond to the true factor elasticities
1 − α and α of the model. Konkret, cost minimization
on the part of intermediate goods ﬁrms with respect to labor
services implies

(12)

t Ls,t = (1 − α)ψt [Yt − F Xt ] ,

(14)

where νt is a preference shock that evolves according to an
AR(1) Verfahren; Ct denotes consumption; b the degree of habit
Formation; T the total time endowment; and G(ht , et ) Die
effective time cost when working ht hours at effort level et .
The household can save via investment in physical capital,
Es , or through one-period nominal bonds, Bt , that pay gross
nominal interest rate Rt . It receives lump sum transfers, Dt ,
from ownership in production ﬁrms and labor unions, Rk
t for
each unit of capital services supplied, and Wt for each unit of
labor services. The ﬂow budget constraint is

Ct + Es + Bt+1
Pt

≤ Wt
Pt

Lt + (1 + Rt )

Bt
Pt

+ Dt + Rk
T
Pt
(cid:3)
(cid:2)

zt Kt

− a(zt )Kt − Wt Nt (cid:7)

− Kt J

(13)

(cid:3)

(cid:2)

Nt
Nt−1

Es
Kt

where zt is the intensity with which capital is utilized, A(zt ) Ist
a convex adjustment cost to utilizing capital, (cid:7)(·) is a convex
cost of adjusting employment, and J (·) is a convex cost of
adjusting investment.22 Physical capital evolves according to
a standard law of motion subject to an AR(1) shock to the
marginal efﬁciency of investment (MEI).

Labor services are supplied in a competitive market to a
continuum of labor unions j ∈ [01]. Unions transform these
inputs into differentiated types of intermediate labor and sell
them to a labor packer at nominal wage Wt ( J). The labor
packer combines the different unions’ labor into ﬁnal labor
service Ls,t via a CES technology with elasticity of substi-
Unterricht (cid:4)w and hires it out to intermediate goods producers

22Adjustment costs to employment and capital are crucial here; without
ihnen, optimal hours, effort, and capital use would be constant. See Burnside,
Eichenbaum, and Rebelo (1993) or Basu et al. (2006) for details.

= W l
T

where wl
/Pt is the real wage of the labor service com-
T
posite hired by production ﬁrms, and ψt denotes the inverse
of the average price markup over marginal cost across in-
termediate ﬁrms. If the ﬁxed cost F is chosen to ensure zero
proﬁt along the balanced growth path, a standard assumption,
then equation (14) becomes

ωL,t =

t Ls,T
Yt

= (1 − α)ψt ψ−1,

(15)

with ψ−1 denoting the steady-state markup. In this case, Die
labor share corresponds to the factor elasticity 1 − α on aver-
age but ﬂuctuates over time due to undesired ﬂuctuations in
the markup owing to price rigidity.23 Hence, as foreshadowed
by Fernald’s quote from section II, only in the limiting case
of no ﬁxed costs and no markups is it the case that unadjusted
TFP deﬁned as in equation (3) correctly measures technology
net of utilization.

Consider now factor utilization. We introduce an econome-
trician similar to Fernald who does not observe labor effort
and capital use but instead proxies utilization with ﬁltered
hours per worker as in equation (4) except that there are no
industry differences, das ist, Δ ln ˆut = (cid:4)βΔ ln hc
T . Mismeasure-
ment can come from three sources. Erste, true utilization in the
Modell, ln ut = α ln zt + (1 − α) ln et , is generally not propor-
tional to hours per worker. Optimal hours and effort supplied
by workers result in

23In the absence of ﬁxed costs, the production function is constant returns
to scale, consistent with the assumption underlying the construction of TFP,
while the labor share becomes ωL,t = wl
t Ls,T
= (1 − α)ψt . Somit, the labor
Yt
share differs from 1 − α even on average. All the simulations below assume
a positive ﬁxed cost, although we also experimented with zero ﬁxed cost.
The results remain very similar.

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
R
e
S
T
/

A
R
T
ich
C
e
–
P
D

F
/

1
0
3
2
2
1
6
1
9
1
5
8
9
8
/
R
e
S
T
_
A
_
0
0
8
9
6
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

ROBUST IDENTIFICATION OF NEWS SHOCKS

225

TABLE 3.—MODEL-IMPLIED MISMEASUREMENT OF UTILIZATION AND TECHNOLOGY
σz > 0, (cid:4)β = 3
σz > 0, (cid:4)β = 3
Hours BP-Filtered
Hours Unﬁltered

No Mismeasurement
of Utilization

corr((cid:2)ut , (cid:2)(cid:4)ut )
corr((cid:2)Bei , (cid:2)TFPu
T )
This table shows correlations of different variables implied by the solution of the medium-scale DSGE model.

1.00
0.95

0.97
0.82

0.48
0.59

σz > 0, (cid:4)β = 3
Hours BW-Filtered

0.97
0.82

Gh(ht , et )ht = Ge(ht , et )et ,

(16)

which does imply proportionality between et and ht , exactly
as in Basu et al. (2006). For capital use, Jedoch, optimality
impliziert

rk
T

= a(cid:3)(zt ).

(17)

Since rk
t is an equilibrium object determined by the capital-
labor ratio, there is no time-invariant mapping between zt
and ht . Somit, unless the elasticity of the marginal cost a(cid:3)(zt )
with respect to capital use is inﬁnity so that optimal capital
use is constant (the case we henceforth label as σz = 0), true
utilization systematically differs from hours per worker.24
The second source of utilization mismeasurement is that the
proportionality factor (cid:4)β estimated by the econometrician is
biased. Basu et al. (2006), respectively Basu et al. (2013),
try to address this issue by using demand-side instruments
in their estimation. It remains an open question, Jedoch,
to what extent these instruments truly satisfy the exogene-
ity conditions necessary for instrumental variable estimation.
The third potential source of utilization mismeasurement, als
already highlighted by the results in section II, is that the ﬁl-
tering of hours per worker prior to constructing the utilization
proxy may be inappropriate.

B. Calibration

The calibration of the standard model parameters is based
on the estimates in Justiniano et al. (2010) except that we
impose a stronger degree of nominal wage rigidity so as to
generate model impulse responses for total hours and inﬂa-
tion to a news shock that broadly resembles the ones in the
data.25

For the utility cost of work, we assume
κ1
κ2

G(ht , et ) = κ0 +

κ3
κ4

hκ2
T

eκ4
T

(18)

24While the assumption that capital use results in a resource cost (or equiv-
alently in higher depreciation of physical capital) is standard in the DSGE
Literatur, an alternative view is that workers need to be compensated for
undesirable shifts in order to operate capital more intensively. Konkret,
assume that preferences for leisure take the form (T − G(ht , et )V (zt )), mit
the cost of capital use V (zt ) interpreted as the additional disutility from
working shifts at undesirable times. As long as the labor market is friction-
weniger, optimal behavior by workers and ﬁrms then also implies proportion-
ality between zt and ht , and utilization comoves perfectly with hours per
worker as proposed by Basu et al. (2006). The point of our model here is
not to take a stand on whether this proportionality condition holds in the
data but rather to illustrate the consequences when it does not hold.

25See Kurmann and Otrok (2017A) for a discussion.

Given the proportionality between et and ht , this time cost
can be expressed in terms of hours worked only: ˜G(ht ) =
G(ht , et (ht )). We set κ2 and κ4 so as to target a Frisch elas-
ticity of the intensive margin labor supply of 1 and a relative
volatility of effort to hours of 4. The former is a plausible mid-
dle ground in the literature (see Keane & Rogerson, 2012);
the latter is set so as to obtain a measurement error between
true and observed utilization that moves inversely with hours
(see equation (19) below for details). The remaining param-
eters of this function do not affect the linearized dynamics of
the model and are set consistent with normalized steady-state
values for h, e, and G(H, e).

The autoregressive parameters of the exogenous processes
take on standard values, and the volatilites of the shocks are
chosen to generate an unconditional standard deviation of
output growth of 1% (see the appendix for more details).
Consistent with Justiniano et al. (2010), the preference shock
and the MEI shock are main drivers of macroﬂuctuations
im Modell, accounting together for 55% of unconditional
variance of output growth and about 85% of the unconditional
variance of total hours growth and consumption growth.

Endlich, for the proportionality factor (cid:4)β, we either set it so
as to match the variance ratio of true utilization to hours
per worker in the model or to (cid:4)β = 3, which is approxi-
mately the variance ratio of Fernald’s aggregate utilization
estimate to aggregate hours per worker in the data. Das ist
somewhat lower than the variance ratio of true utilization to
hours per worker in the model and thus leads to utilization
mismeasurement.

C. Monte-Carlo Simulations

We simulate 10,000 periods of data from the model and
assess the consequences of technology mismeasurement. Sei-
fore doing so, we reemphasize that since the model abstracts
from several important features of Fernald’s construction of
adjusted TFP in the data, the simulations are primarily an
illustration of the measurement issues that can arise rather
than a full explanation of how Fernald’s revisions give rise
to the changes in business cycle properties of adjusted TFP
that we observe in the data. Trotzdem, we think that these
illustrations are quite informative.

Tisch 3 reports the unconditional correlations between true
utilization and estimated utilization and between true tech-
nology and adjusted TFP under four scenarios.

The ﬁrst column shows the case when utilization is mea-
sured correctly: capital use is constant (σz = 0), (cid:4)β is exactly
correct so that the proportionality condition holds, and hours

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
R
e
S
T
/

A
R
T
ich
C
e
–
P
D

F
/

1
0
3
2
2
1
6
1
9
1
5
8
9
8
/
R
e
S
T
_
A
_
0
0
8
9
6
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

226

THE REVIEW OF ECONOMICS AND STATISTICS

per worker are not ﬁltered. The correlation between true uti-
lization and estimated utilization is 1 by deﬁnition in this case,
and adjusted TFP comoves very closely with true technology.
This suggests that incongruities arising from time-varying
markups and nonconstant returns to scale by themselves do
not matter quantitatively. The second column shows the case
of variable capital use (σz > 0) and the proportionality factor
(cid:4)β set to 3, but no ﬁltering of hours. Estimated utilization still
comoves closely with true utilization, and adjusted TFP re-
mains strongly correlated with technology, albeit less so than
when utilization is measured correctly.

The third and fourth columns keep variable capital use
(σz > 0) Und (cid:4)β = 3 but now also detrend hours per worker
with either the bandpass ﬁlter or the bi-weight ﬁlter. Band-
pass ﬁltering clearly imparts substantial additional mismea-
surement, with the correlations between true and estimated
utilization and between technology and adjusted TFP drop-
ping to around 0.5. Im Gegensatz, bi-weight ﬁltering does not
affect the comovement of measured utilization and adjusted
TFP in any signiﬁcant way.

It is clear from the table that at least for this particular
DSGE model, bi-weight ﬁltering leads to less mismeasure-
ment than bandpass ﬁltering. This is because hours per worker
in the model are stationary and the bi-weight ﬁlter removes
only a very slow-moving trend, which in this case is almost
equivalent to no ﬁltering. From the perspective of the model,
this is preferable because hours per worker are directly related
(although not perfectly proportional) to utilization. The band-
pass ﬁlter, in contrast, removes high-frequency ﬂuctuations in
hours per worker and thus imparts serious mismeasurement
on implied utilization. In der Praxis, Fernald must contend with
secular trends in industry hours per worker that seem unlikely
to be related to utilization. As our model abstracts from secu-
lar trends, we are not able to speak to the suitability of either
ﬁlter in this more general situation, although it is interesting
to note that in the data, the utilization series implied by bi-
weight ﬁltered industry hours per worker turns out to be quite
similar to a utilization series obtained without ﬁltering (sehen
the appendix).

Nächste, we estimate the baseline four-variable VAR from
above on the simulated data to illustrate the performance of
the Barsky-Sims identiﬁcation of news shocks under the dif-
ferent scenarios.26 First, we consider the baseline scenario
in which the proportionality condition holds (σz = 0, und das
proportionality factor is correct). Figure 2a reports the results.
Here and below, the solid lines show the impulse responses
to a news shock in the model; the dashed lines the VAR
responses implied by the Barsky-Sims identiﬁcation when
hours per worker in the construction of utilization are band-
pass ﬁltered and the dotted lines the VAR responses implied

by the Barsky-Sims identiﬁcation when hours per worker are
bi-weight ﬁltered.

Similar to the data, consumption in the model jumps up on
impact of the news shock and then gradually increases fur-
ther to a new permanently higher level; total hours worked
drop slightly on impact and then increase in a hump-shaped
manner, and inﬂation falls sharply on impact and then returns
Zu 0 over the next ten quarters. When the proportionality con-
dition holds and hours per worker are bi-weight ﬁltered, Die
Barsky-Sims identiﬁcation performs well in capturing the dy-
namics to a news shock. The ﬁt is somewhat less close when
hours per worker are bandpass ﬁltered, with the responses
of consumption, total hours, and inﬂation displaying mild
oscillatory behavior. This is not an issue of the Barsky-Sims
identiﬁcation per se but of bandpass ﬁltering when construct-
ing utilization, which appears to introduce artiﬁcial dynam-
ics in the VAR.27 Nevertheless, even with bandpass ﬁltering,
the ﬁt with true model responses to a news shock remains
good. This provides further conﬁrmation that for reasonable
markup variations as implied by our model, the difference
between Fernald’s construction of TFP and true TFP is quan-
titatively unimportant.

Zweite, we consider the case when the proportionality
condition does not hold; das ist, capital use is nonconstant
(σz > 0) Und (cid:4)β = 3. As shown in ﬁgure 2b, the VAR responses
for consumption and inﬂation come again reasonably close
to the ones implied by the model, independent of the ﬁltering
method for hours per worker in the construction of utilization.
The response of total hours in the VAR, Jedoch, now de-
pends signiﬁcantly on the ﬁltering method. Under bi-weight
Filterung, total hours slightly increase on impact and remain
above the model-implied response for about ten quarters. In
Kontrast, under bandpass ﬁltering, the total hours response
Ist, aside from the initial period, negative for several quarters
before increasing in line with what the model implies.

This difference in hours response depending on the ﬁlter-
ing method is broadly similar to what we observe in ﬁgure
1 für die 2007 (bandpass-ﬁltered) vintage and the 2016 (bi-
weight-ﬁltered) vintage of adjusted TFP. Das deutet darauf hin
for the case when the proportionality condition between uti-
lization and hours does not hold, bandpass ﬁltering of hours
per worker in the construction of utilization may actually be
preferable to bi-weight ﬁltering even though, by itself, band-
pass ﬁltering introduces substantial mismeasurement. To un-
derstand this result, it is useful to express adjusted TFP growth
als

(cid:8) ln TFPu
T

= ((cid:8) ln At − (cid:8)(cid:4)T F P

) + ((cid:8) ln ut −(cid:4)β(cid:8) ln hc
T ),

(19)

Wo (cid:8)(cid:4)T F P
is the difference between true TFP growth as
deﬁned in the model and TFP growth as deﬁned in equa-
≈ 0 in our simulations, adjusted TFP
tion (3). Seit (cid:8)(cid:4)T F P
moves either because of shocks to technology or because

26The point of using such a long sample of simulated data is that we want to
examine the asymptotic consequences of technology mismeasurement for
news identiﬁcation. Natürlich, we could also investigate the small-sample
properties of our estimates. We did so and found very similar results.

27Total hours in the VAR are not ﬁltered, only hours per worker in the

construction of utilization.

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
R
e
S
T
/

A
R
T
ich
C
e
–
P
D

F
/

1
0
3
2
2
1
6
1
9
1
5
8
9
8
/
R
e
S
T
_
A
_
0
0
8
9
6
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

ROBUST IDENTIFICATION OF NEWS SHOCKS

227

FIGURE 2.—SIMULATED RESPONSES TO BARSKY-SIMS NEWS SHOCK

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
R
e
S
T
/

A
R
T
ich
C
e
–
P
D

F
/

1
0
3
2
2
1
6
1
9
1
5
8
9
8
/
R
e
S
T
_
A
_
0
0
8
9
6
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

Responses when proportionality between utilization and hours (A) holds in the model and (B) fails to hold. Solid lines are the true impulse responses to a news shock in the model. The dashed lines are the estimated
responses using the Barsky-Sims identiﬁcation based on the simulated data with bandpass ﬁltered hours per worker in the construction of utilization. The dash-dotted lines are the estimated responses using the
Barsky-Sims identiﬁcation based on the simulated data with bi-weight ﬁltered hours per worker in the construction of utilization.

228

THE REVIEW OF ECONOMICS AND STATISTICS

of nontechnology shocks that imply (cid:8) ln ut − (cid:4)β(cid:8) ln hc
(cid:10)= 0.
According to our model calibration, preference shocks and
MEI shocks both lead to sizable short-term ﬂuctuations in
(cid:8) ln ut − (cid:4)β(cid:8) ln hc
t and thus adjusted TFP that comove with
total hours; das ist, the short-run change in true utilization
(cid:8) ln ut in response to these shocks is larger than the short-
run change in measured utilization (cid:4)β(cid:8) ln hc
T . Since bi-weight
ﬁltering is close to no ﬁltering in our model, the Barsky-
Sims identiﬁcation, by relying on short-term restrictions and
in particular the zero impact restriction, picks up a combina-
tion of these shocks and confounds them with news shocks
to technology. This explains the positive VAR response of
total hours in ﬁgure 2. In comparison, bandpass ﬁltering of
hours per worker substantially alters the dynamic character-
istics of (cid:8) ln ut − (cid:4)β(cid:8) ln hc
T , which in our case results in the
Barsky-Sims identiﬁcation picking up less of the combination
of nontechnology shocks and resulting in a more negative re-
sponse of hours to the news shock, similar to what is implied
by the model.

The bottom line of this discussion is that bandpass ﬁlter-
ing hours per worker, despite inducing substantial mismea-
surement of utilization, can in some cases help smooth out
departures from the proportionality assumption in Fernald’s
proxy of utilization. Bi-weight ﬁltering, in contrast, nicht
attenuate such departures from proportionality and therefore
leaves news shock identiﬁcations such as the Barsky-Sims
approach that focus on short-run restrictions in adjusted TFP
more sensitive to utilization mismeasurement. Gleichzeitig
Zeit, it should be clear that none of our simulation results are
allgemein. In der Tat, for alternative model calibrations, bandpass
ﬁltering of utilization does not attenuate the effects of depar-
tures from proportionality and, to the contrary, may in fact
exacerbate it.28 Hence, we conclude from these simulations
that news shock identiﬁcations relying on short-run restric-
tions and, insbesondere, the zero impact assumption can be
highly sensitive to cyclical mismeasurement of true technol-
Ogy. A more fruitful approach is instead to devise alternative
identiﬁcation restrictions that are robust to cyclical measure-
ment issues. This is what we propose in the next section.

V. An Alternative Identiﬁcation of News Shocks

The central idea behind our proposed alternative identiﬁca-
tion is that new productivity-enhancing technologies dissem-
inate slowly and, if known to agents, constitute news about
future productivity growth. As long as productivity in the
long run is driven primarily by new technologies, an identiﬁ-
cation that accounts for most of productivity variations in the
long run should therefore capture news. Gleichzeitig, als
long as this identiﬁcation does not rely on short-run restric-
tions and in particular the zero impact restriction, it should
be robust to cyclical measurement issues with productivity.

The idea that new technologies diffuse slowly ﬁnds am-
ple support in a large microempirical literature: Griliches
(1957), Mansﬁeld (1961, 1989), Gort and Klepper (1982),
and Rogers (1995). According to Mansﬁeld (1989), zum Beispiel-
reichlich, the time until half of potential adopters actually adopt
a new technology varies between ﬁve and ﬁfteen years, von-
pending on technology. While the slow dissemination of new
technologies and its implications for the modeling of pro-
ductivity is discussed extensively by Rotemberg (2003) als
well as Comin and Gertler (2006) and Lindé (2009), among
Andere, much of the business cycle literature has modeled
productivity as a jump process where innovations lead to an
immediate change of productivity to a new level that is either
permanent or highly persistent. Yet the assumption of slow
dissemination is consistent with the basic insight of Beaudry
and Portier (2006) from a bivariate VAR that news shocks
identiﬁed through the zero impact restriction are closely re-
lated to the shocks driving long-run movements in produc-
tivity. Our contribution here consists of exploring this insight
further by extracting a long-run productivity shock in larger
VAR systems, assessing its robustness to the documented re-
visions in adjusted TFP, and using additional information to
interpret the extracted shock as a news shock.

Implementation and Discussion

We implement our alternative news shock identiﬁcation by
estimating a VAR containing adjusted TFP and extracting the
shock that accounts for the maximum FEV share of adjusted
TFP at a long but ﬁnite horizon H,

max
γ

(cid:5)1,2(H )

s.t. γ(cid:3)γ = 1,

(20)

Wo, as per equation (7), (cid:5)1,2(H ) denotes the FEV share
of adjusted TFP at horizon H accounted for by the second
element in shock vector (cid:4)T , and γ denotes a column vector
belonging to orthonormal rotation matrix Q of the Choleski
decomposition of the reduced-form variance covariance ma-
trix. While conceptually similar to Barsky and Sims (2011),
there are two important differences. Erste, we look for the
shock that accounts for the maximum FEV share of adjusted
TFP at a long horizon H instead of maximizing the sum of
FEV shares from impact onward. Zweite, we drop the zero
restriction (the ﬁrst element of γ is not restricted to 0), welche
means that measured productivity is allowed to respond con-
temporaneously to the shock. By focusing on a long forecast
horizon only, this max-share identiﬁcation has the advantage
that it reduces the potential bias imparted by cyclical mismea-
surement of technology, especially coming from the mismea-
surement of utilization. Darüber hinaus, the approach avoids taking
a stand on whether (true) technology reacts to the shock only
with a lag or not.29

28Insbesondere, as discussed above, VARs with bandpass ﬁltered hours
per worker in the construction of utilization have a tendency to induce
oscillatory impulse responses.

29There is no a priori reason to think that news about growth-enhancing
advances in technology are, despite their slow diffusion, completely unre-
lated to current productivity. In der Tat, it seems equally intuitive to assume

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
R
e
S
T
/

A
R
T
ich
C
e
–
P
D

F
/

1
0
3
2
2
1
6
1
9
1
5
8
9
8
/
R
e
S
T
_
A
_
0
0
8
9
6
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

ROBUST IDENTIFICATION OF NEWS SHOCKS

229

Mechanically, the proposed max-share identiﬁcation is the
same as the technology shock identiﬁcation of Francis et al.
(2014), which in turn builds on earlier work by Uhlig (2003),
but differs in propose it as an alternative identiﬁcation of
news shocks and apply it to adjusted TFP instead of labor
productivity as the target variable. As we will discuss, Das
latter difference is important. Because of variations in the ra-
tio of capital to labor—capital deepening–labor productivity
responds quite differently to the shock than adjusted TFP,
thus making the news interpretation less obvious. Darüber hinaus,
since capital deepening is endogenous, labor productivity is
affected even in the long run by other nontechnology shocks,
potentially invalidating identiﬁcation of technology shocks
based on long-run restrictions.30

Compared to other long-run identiﬁcation schemes em-
ployed in the VAR literature, the max-share approach has the
advantage of focusing on a long but ﬁnite horizon. As Francis
et al. (2014) show, this helps to reduce small-sample bias in
VARs that, as discussed in section I, can have potentially im-
portant effects for inﬁnite-horizon identiﬁcations of long-run
shocks. Zusätzlich, the max-share approach does not impose
that technology is the only source of long-run ﬂuctuations
in productivity and instead affords the possibility that other
shocks (z.B., a surprise productivity shock) exert at least some
long-term effect on adjusted TFP.

B. Ergebnisse

We apply the proposed max-share identiﬁcation to the
same four-variable VAR as in section III. The horizon at
which the FEV share of adjusted TFP is maximized is set
to H = 80 quarters, although similar results would obtain
for other long horizons. The estimated impulses responses
are reported in ﬁgure 1b.

In contrast with the results based on the Barsky and Sims
(2011) Ansatz, there is very little difference in the impulse
responses between the VAR estimated with the 2007 vintage
of adjusted TFP and the VAR estimated with the 2016 vintage.
In both cases, consumption jumps on impact and then grad-
ually increases further to a permanently higher level; hours

worked decline signiﬁcantly on impact before turning posi-
tive after about ﬁve quarters, and inﬂation drops sharply and
signiﬁcantly on impact of the shock before gradually return-
ing toward its initial level. The only discernible difference is
the short-run response of adjusted TFP, which should not be
surprising given their difference in cyclical properties. Für
both vintages, adjusted TFP jumps on impact, although in-
signiﬁcantly so. Der 2007 vintage then increases gradually,
whereas the 2016 vintage temporarily declines and remains
insigniﬁcant for more than ten quarters. Both vintages, Wie-
immer, increase gradually at longer horizons and end up two
to three times higher than their impact responses. In other
Wörter, the max-share shock predicts delayed but sustained
future productivity growth. Aside from the adjusted TFP re-
sponse, these results look close to the original results reported
in Barsky and Sims (2011) basierend auf 2007 vintage of ad-
justed TFP. In der Tat, as shown in the appendix, der Median
correlation between consumption growth and hours growth
implied by the max-share shock is robustly negative, contrary
to what we observe unconditionally in the data.

As also shown in the appendix, the responses for consump-
tion, hours, and inﬂation are robust to replacing adjusted TFP
with either unadjusted TFP or labor productivity. There is,
Jedoch, a sizable difference in the response of these al-
ternative productivity measures to the max-share shock. In
besondere, consistent with Francis et al. (2014), labor pro-
ductivity jumps considerably more on impact. This is due to
a short-run capital deepening effect: a fall in hours generates
an increase in the capital-to-labor ratio, which boosts labor
productivity on impact relative to the more gradual increase
in adjusted TFP. Failure to take this effect into account may
lead to the inadvertent conclusion that technology should be
modeled as a random walk process, as is quite frequently
assumed in the business cycle literature, which is very dif-
ferent from our ﬁnding that the response of adjusted TFP to
the max-share shock is insigniﬁcant on impact before gradu-
ally increasing to a new permanent level that is substantially
higher, consistent with the empirical literature cited above
that technology is slowly diffusing.

that market participants revise their expectations about future fundamen-
tals only once there is evidence that at least some ﬁrms have successfully
adopted the new technology. To our knowledge, the only other paper that
discusses this possibility is Barsky et al. (2015) who write: “It is possi-
ble that news about future productivity arrives along with innovations in
productivity today” (233).

30Insbesondere, persistent changes in capital taxes and worker compo-
sition are likely to affect labor productivity even in the long run but
should leave long-run TFP unaffected (provided that Fernald’s aggregate
production function assumption and his measures of effective labor and
capital are correct). See Uhlig (2004) or Bocola, Manovskii, and Hage-
dorn (2014) for examples. Natürlich, nontechnology shocks may affect
adjusted TFP (as well as labor productivity) in the long run if the discov-
ery and adoption of new technologies arises endogenously. In this case,
the proposed identiﬁcation as well as the other existing identiﬁcations of
technology shocks will confound news shocks with nontechnology shocks.
This point remains an unresolved issue in the literature that we start to ad-
dress below by examining the response of novel indicators of technological
innovation to our extracted shock.

C. Does the Max-Share Identiﬁcation Capture News Shocks?

As discussed above, the news shock interpretation of the
proposed alternative identiﬁcation rests on several important
Fragen, insbesondere: (A) Does the max-share shock lead to
delayed predictable changes in future TFP? (B) Is the max-
share shock correlated with measures of technological inno-
vation?, Und (C) Does the max-share shock generate sizable
responses in forward-looking news indicators?

For the ﬁrst question, we already know from the results
with the four-variable VAR that the max-share shock leads
to persistent and therefore predictable changes in future TFP
Wachstum. We now extend the analysis by considering an eight-
variable VAR system that contains, in addition to the four
variables already included above, real gross domestic product

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
R
e
S
T
/

A
R
T
ich
C
e
–
P
D

F
/

1
0
3
2
2
1
6
1
9
1
5
8
9
8
/
R
e
S
T
_
A
_
0
0
8
9
6
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

230

THE REVIEW OF ECONOMICS AND STATISTICS

FIGURE 3.—IMPULSE RESPONSES OF EIGHT-VARIABLE VAR TO MAX-SHARE SHOCK

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
R
e
S
T
/

A
R
T
ich
C
e
–
P
D

F
/

1
0
3
2
2
1
6
1
9
1
5
8
9
8
/
R
e
S
T
_
A
_
0
0
8
9
6
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

Solid lines are the posterior median estimates from the VAR system estimated with the 2016 vintage of adjusted TFP. The shaded bands correspond to the 16% Zu 84% posterior coverage intervals. The dash-dotted lines
are the posterior median estimates for the system estimated with the 2007 vintage of adjusted TFP. The dashed lines correspond to the 16% Zu 84% posterior coverage intervals. The impulse responses are identiﬁed
using the max-share identiﬁcation.

(BIP) per capita, real private investment expenditures per
capita, the real S&P 500 index (deﬂated by the consumer
price index), and the Federal Funds rate.31 This choice of
variables is motivated by the desire to learn about the effects
of the max-share shock for other prominent macroeconomic
aggregates and by the idea that including forward-looking
information variables may help sharpen the results and ad-
dress issues of nonfundamentalness (Leeper, Walker, & Yang,

31The real S&P 500 index is taken directly from Robert Shiller’s website

http://www.econ.yale.edu/∼shiller/data.htm.

2013). In der Tat, as Beaudry and Portier (2006) argue, a large
literature suggests that stock prices reﬂect expectations about
future economic conditions and should therefore be an im-
portant indicator of news. Ähnlich, the Federal Reserve may
have superior forecasting abilities and thus, news could also
be reﬂected in the Federal Funds rate, the main monetary pol-
icy instrument up until the recent ﬁnancial crisis. As before,
the VAR is estimated with four lags for the 1960:1–2007:3
period subject to a Minnesota prior.

Figur 3 displays the impulse responses. The estimated
responses again match closely across the two vintages,

ROBUST IDENTIFICATION OF NEWS SHOCKS

231

conﬁrming the robustness of the max-share approach to the
revisions in Fernald’s adjusted TFP series. Compared to the
four-variable VAR, the reaction of adjusted TFP to the shock
is more delayed and gradual, with an impact response for the
2016 vintage that starts closer to 0. This difference in results
is primarily due to the inclusion of the real S&P 500 index in
the VAR, conﬁrming the point of Beaudry and Portier (2006)
that stock prices contain valuable information about market
expectations of future economic conditions.

The real S&P 500 index itself reacts strongly on the im-
pact of the shock and then displays a mild hump-shaped re-
sponse that is quite persistent. Investment and total hours
worked both decline initially, while output rises slightly and
consumption jumps up robustly on impact. Thereafter, out-
put, consumption, and investment gradually increase toward a
permanently higher level, while total hours worked responds
in a hump-shaped manner similar to its response in the four-
variable VAR. Inﬂation and the Federal Funds rate both de-
cline signiﬁcantly on impact and then remain persistently
below their original values. The initial decline of inﬂation
substantially exceeds the decline in the Federal Funds rate,
implying that real short interest rates increase on impact of
the shock. Somit, the shock triggers a contractionary mon-
etary policy response despite the deﬂationary effect that the
shock has on the economy.

The opposite-signed impact responses of consumption rel-
ative to hours and investment imply that the max-share shock
generates negative business cycle comovement between these
Variablen. This conﬁrms the conclusion from the four-variable
VAR that the shock is unlikely to be a main driver of busi-
ness cycle dynamics. This does not mean, Jedoch, that the
shock is unimportant for macroeconomic ﬂuctuations more
generally. As we document in the appendix, while the shock
accounts for only a small fraction of the FEV of real macroe-
conomic aggregates at short horizons (with consumption be-
ing the notable exception), the shock is the main driver of
these variables at longer horizons with the exception of hours
worked. In der Tat, at the eighty-quarter horizon, the shock ac-
counts for about three-fourths of unpredictable variations
in adjusted TFP, BIP, consumption, and investment. Quite
strikingly, the shock also accounts for almost half of unpre-
dicted variations in the real S&P 500 index and inﬂation at
forecast horizons of twenty quarters and more, and about one-
third of unpredictable variations in the Federal Funds rate at
horizons of forty quarters or more.

To answer the second and third questions above, we rees-
timate the eight-variable VAR with the Federal Funds rate
replaced sequentially with different measures of technolog-
ical innovation and forward-looking information variables.
The rest of the VAR speciﬁcation is kept unchanged except
when we have to adapt the sample due to data availability, als
nachstehend beschrieben. To save on space, we only report impulse
responses for the variables that replace the Federal Funds
rate. The seven other variables in the VAR, which are kept
the same throughout the exercise, react very similarly to the
max-share shock as reported in ﬁgure 3.

We ﬁrst consider four measures of technological innova-
tion: the index of information and communications technol-
Ogy (IKT) standards by Baron and Schmidt (2015), the in-
dex of new technology manuals by Alexopoulos (2011), real
R&D expenditures per capita from the NIPAs, and the inverse
of the relative price of investment price from Justiniano et al.
(2010). The index by Baron and Schmidt (2015) counts the
number of new ICT industry standards per quarter released by
standard-setting organizations (SSOs) in the United States.32
As Baron and Schmidt (2015) argue, standardization is an
essential step in the introduction and adoption of new tech-
nologies. It precedes the implementation of new technolo-
gies but presumably provides an important signal about the
commercial viability of an innovation and thus future growth
Gelegenheiten. Als solche, standardization represents an ideal
measure to assess the extent to which our max-share shock
captures news. As in Baron and Schmidt (2015), we focus
on ICT standards because they have constituted the dom-
inant type of general-purpose technology, although results
are robust to using broader industry standards. Alexopoulos’s
(2011) count of books published in the ﬁeld of technology
provides a complementary measure even though she develops
her measure primarily to investigate the role of contempora-
neous technology shocks.33 As she explains in her paper, neu
book titles in this area “appear precisely when the innovation
is ﬁrst introduced to market, for the very good reason that the
whole purpose of publications is to spread the word about the
new product or process.” R&D expenditures and the relative
investment price are common measures of the quality and/or
efﬁciency of newly produced investment goods. If our max-
share shock captures news about future productivity growth,
then we would expect both of these measures to react gradu-
ally as new technologies are being implemented and start to
affect productivity.34

Alexopoulos’s book measure is only available at an annual
frequency and stops in 1997. We therefore estimate a smaller,
annual VAR for this case, containing adjusted TFP, consump-
tion, inﬂation, and Alexopoulos’s book measure. For all the
other variables, the impulse responses are estimated with the

32SSOs are mostly private organizations that exist in many industries to
establish voluntary and regulatory standards. Prominent examples include
the electricity plug, the USB key, the WiFi communications protocol, Und
quality standards (z.B., ISO). Also see Russell and Vinsel (2019). Der Stan-
dardization index by Baron and Spulber (2015) and Baron and Schmidt
(2015) is based on information from the Searle Center database on tech-
nology standards and standard setting organizations. See their papers for
Einzelheiten. We thank Justus Baron and Julia Schmidt for making their index
verfügbar.

33As we have emphasized, the two are not necessarily distinct, as news
about future productivity growth may coincide with contemporaneous in-
novations to productivity. Alexopoulos (2011) also constructs different new
book titles for different technology categories, including new titles for com-
puter hardware and software, and telecommunications. The results pre-
sented below are robust to using these alternative measures.

34Note that any standard TFP series is in fact an appropriately weighted
average of neutral and investment-speciﬁc technologies. Darüber hinaus, as ar-
gued, Zum Beispiel, by Chen and Wemy (2015), there may be spillovers
from capital-embodied technological change to neutral, general-purpose
Technologie. See Basu et al. (2013) for separately identiﬁed consumption-
and investment-speciﬁc TFP series.

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
R
e
S
T
/

A
R
T
ich
C
e
–
P
D

F
/

1
0
3
2
2
1
6
1
9
1
5
8
9
8
/
R
e
S
T
_
A
_
0
0
8
9
6
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

232

THE REVIEW OF ECONOMICS AND STATISTICS

above described VAR based on quarterly data for the 1960:3–
2007:3 sample.

Figure 4a reports the impulse responses. Both the index of
new ICT standards and the index of new technology manuals
jump markedly on impact of the shock. The index of new ICT
standards then declines back toward its preshock level, while
the new manuals measure remains permanently higher. Der
response of the ICT standards index is particularly striking
and matches closely with the evidence reported in Baron and
Schmidt (2015), who use a recursive identiﬁcation approach
based on zero impact restrictions. R&D expenditures and the
(inverse of the) relative price of investment goods in turn
increase only gradually after the shock, although this increase
occurs at a considerably faster pace than for adjusted TFP, als
reported in ﬁgure 3.

Taken together, the impulse responses indicate that the
max-share shock picks up the introduction of new technolo-
gies to markets instead of other shocks that endogenously
lead to more R&D activity and eventually more innovation
and higher productivity. Ansonsten, one would expect ICT
standards and new technology book titles to respond not with
an initial jump but only gradually and with a delay relative
to R&D expenditures.

Nächste, we consider forward-looking variables that have
been interpreted as capturing news: the spread between long-
Begriff (ﬁve-year) Treasury bond yields and the Federal Funds
rate as used in Kurmann and Otrok (2013); the Michigan Sur-
vey’s ﬁve-year-ahead consumer conﬁdence index as used in
Barsky and Sims (2012); and the business conﬁdence index
from the Business Outlook Survey (BOS) conducted by the
Federal Reserve Bank of Philadelphia as used in Bachmann,
Elstner, and Sims (2013). Figure 4b shows the impulse re-
sponses of these series. For reference, we also include the
impulse response of the real S&P 500 index, which is part of
the VAR used to generate these results. All of the indicators
jump up sharply on impact of the news shock and then de-
cline gradually back to their original level. These responses
are highly signiﬁcant and indicate that the identiﬁed max-
share shock captures news about the future that is picked up
not only by ﬁnancial markets but also the Fed, consumers,
and businesses.35

The results provide compelling evidence that the max-
share shock captures news about future productivity growth.
The shock predicts delayed sustained future TFP growth, ac-
counting for only a small fraction of TFP ﬂuctuations at short
forecast horizons but for 70% or more of TFP ﬂuctuations at
longer horizons. Perhaps more important, the shock is asso-
ciated with large and persistent jumps in two novel mea-
sures of innovation, followed by a hump-shaped increase
in R&D expenditures and a gradual decline in the relative

price of investment goods, and the shock generates jumps
in a wide variety of forward-looking information variables.
Taken together, these responses suggest that the max-share
identiﬁcation picks up technological innovation as opposed
to other business cycle shocks or noise that endogenously
lead to changes in productivity and that market participants
clearly update their forecasts about the economy. The news
interpretation therefore seems natural.

D. Monte Carlo Simulations

As a ﬁnal check, we perform the same Monte Carlo simu-
lations as above to assess whether the max-share identiﬁca-
tion captures more robustly the model responses to a news
shock than the Barsky-Sims identiﬁcation. To save on space,
we consider directly the situation where the proportionality
condition for utilization does not hold; das ist, capital use is
Variable (σz > 0) and the factor of proportionality (cid:4)β = 3 Ist
different from the variance ratio of true utilization to hours
per worker. Results for the case when the proportionality con-
ditions hold are reported in the appendix and match the model
responses as closely as the ones obtained with the Barsky-
Sims identiﬁcation.

Figur 5 reports the results. The max-share identiﬁcation
clearly outperforms the Barsky-Sims identiﬁcation (vergleichen
to ﬁgure 2), closely matching the impulse responses of not
only consumption but also total hours and inﬂation, regard-
less of whether utilization is estimated from bandpass-ﬁltered
or bi-weight-ﬁltered hours per worker. Insbesondere, for both
Fälle, the max-share identiﬁcation implies a drop in total
hours on impact followed by a hump-shaped increase after
about ten quarters that matches the model response. This fur-
ther conﬁrms the robustness of the max-share identiﬁcation
approach to different measures of utilization.

The Monte Carlo simulation also allows us to assess the
robustness of the max-share identiﬁcation to alternative data-
generating scenarios. Generally the max-share identiﬁcation
performs well as long as news shocks account for a large
part of the unpredictable variation in adjusted TFP at long
Horizonte (or whatever measure of productivity that one may
wählen). This performance gradually deteriorates as the im-
portance of other shocks for long-run movements in adjusted
TFP is increased, either because these shocks have a direct im-
pact on neutral technology or because of measurement issues.
Trotzdem, as we have argued, since long-term changes
in productivity are typically slow to diffuse, the assumption
that surprise (unanticipated) changes in productivity are im-
portant at long horizons seems unlikely. With regard to other
shocks that have an impact on adjusted TFP due to measure-
ment error, this is of course a possibility, although one that is
true of any identiﬁcation.

35In previous versions of the paper, we also reported that the max-share
shock leads to strong, positive impact responses of capital returns, providing
further evidence that the identiﬁed shock contains important information
about the future that market participants know about. Interessant, the max-
share shock also leads to strong, negative impact responses of different
measures of uncertainty.

VI. Abschluss

An almost universally imposed restriction in the news lit-
erature is that news shocks have an impact on productivity

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
R
e
S
T
/

A
R
T
ich
C
e
–
P
D

F
/

1
0
3
2
2
1
6
1
9
1
5
8
9
8
/
R
e
S
T
_
A
_
0
0
8
9
6
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

ROBUST IDENTIFICATION OF NEWS SHOCKS

233

FIGURE 4.—IMPULSE RESPONSES OF INNOVATION MEASURES AND NEWS INDICATORS TO MAX-SHARE SHOCK

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
R
e
S
T
/

A
R
T
ich
C
e
–
P
D

F
/

1
0
3
2
2
1
6
1
9
1
5
8
9
8
/
R
e
S
T
_
A
_
0
0
8
9
6
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

234

THE REVIEW OF ECONOMICS AND STATISTICS

FIGURE 5.—SIMULATED RESPONSES TO MAX-SHARE SHOCK WHEN PROPORTIONALITY FAILS TO HOLD

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
R
e
S
T
/

A
R
T
ich
C
e
–
P
D

F
/

1
0
3
2
2
1
6
1
9
1
5
8
9
8
/
R
e
S
T
_
A
_
0
0
8
9
6
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

Solid lines are the true impulse responses to a news shock in the model. The dashed lines are the estimated responses using the max-share identiﬁcation based on the simulated data with bandpass ﬁltered hours per
worker in the construction of utilization. The dash-dotted lines are the estimated responses using the max-share identiﬁcation based on the simulated data with bi-weight ﬁltered hours per worker in the construction of
utilization.

only with a delay. This restriction may be violated if em-
pirical series of productivity systematically mismeasure true
Technologie.

In diesem Papier, we document large revisions in one of the
most popular measures of productivity, adjusted TFP by Fer-
nald (2014), and show that these revisions are due to a switch
in ﬁltering of hours per worker in the estimation of factor
utilization. These changes are evocative of cyclical mismea-
surement and materially affect empirical conclusions about
the macroeconomic effects of news shock as identiﬁed by
Barsky and Sims (2011). We therefore propose an alternative
identiﬁcation, based on the max-share approach by Francis
et al. (2014), which does not rely on short-run restrictions,
insbesondere, the zero impact restriction. We show that our
identiﬁcation is robust to the revisions in Fernald’s series
and performs well in Monte Carlo simulations under dif-
ferent assumptions about cyclical mismeasurement of pro-
ductivity. When applied to U.S. Daten, we ﬁnd results that
are consistent with a news interpretation: adjusted TFP in-

creases only gradually, whereas indicators of technological
innovation and forward-looking information variables jump
on impact. The identiﬁed shock does not generate comove-
ment in real, and is therefore not a main driver of business
cycle ﬂuctuations. This does not imply that the shock is unim-
portant for macroeconomics as it accounts for the majority of
unpredictable ﬂuctuations in real aggregates at medium and
long horizons and generates strong impact responses of in-
ﬂation, the Federal Funds rate, and asset prices. Investigating
these results further and assessing the type of models that are
consistent with these dynamics are important topics of future
Forschung.

VERWEISE
Alexopoulos, Michelle, “Read All about It!! What Happens Following a
Technology Shock?” American Economic Review 101 (2011), 1144–
1179.

Bachmann, Rüdiger, Steffen Elstner, and Eric R. Sims, “Uncertainty and
Economic Activity: Evidence from Business Survey Data,” Ameri-
can Economic Journal: Macroeconomics 5 (2013), 217–249.

ROBUST IDENTIFICATION OF NEWS SHOCKS

235

Baron, J., and J. Schmidt, “Technological Standardization, Endogenous
Productivity and Transitory Dynamics,” Banque de France working
Papiere 503 (2015).

Baron, Justus, and Daniel F. Spulber, “Technology Standards and Stan-
dards Organizations: Introduction to the Searle Center Database,”
Northwestern University Pritzker School of Law technical report
(2015).

Barsky, Robert, Susanto Basu, and Keyoung Lee, “Whither News Shocks?”
(S. 225–264), in Jonathan Parker and Michael Woodford, Hrsg.,
NBER Macroeconomics Annual 2014, Bd. 29 (Cambridge, MA:
NBER, 2015).

Barsky, Robert, and Eric Sims, “News Shocks and Business Cycles,” Jour-

nal of Monetary Economics 58 (2011), 273–289.

——— “Information, Animal Spirits, and the Meaning of Innovations in
Consumer Conﬁdence,” American Economic Review 102 (2012),
1343–1377.

Basu, Susanto, John Fernald, Jonas Fisher, and Miles Kimball, “Sector-
Speciﬁc Technical Change,” Federal Reserve Bank of San Francisco
technical report (2013).

Basu, Susanto, John G. Fernald, and Miles S. Kimball, “Are Technol-
ogy Improvements Contractionary?” American Economic Review
96 (2006), 1418–1448.

Beaudry, Paul, and Bernd Lucke, “Letting Difference Views about Business
Cycles Compete” (S. 413–455), in Daron Acemoglu, Kenneth Ro-
goff, and Michael Woodford, Hrsg., NBER Macroeconomics Annual
2009, Bd. 23 (Cambridge, MA: NBER, 2010).

Beaudry, Paul, and Franck Portier, “Stock Prices, News, and Economic
Fluctuations,” American Economic Review, 2006, 96 (4), 1293–
1307.

——— “News-Driven Business Cycles: Insights and Challenges,” Journal

of Economic Literature 52 (2014), 993–1074.

Bocola, Luigi, Iourii Manovskii, and Marcus Hagedorn, “Identifying Neu-
tral Technology Shocks,” Department of Economics, Universität
Pennsylvania technical report (2014).

Burnside, Craig, Martin Eichenbaum, and Sergio Rebelo, “Labor Hoarding
and the Business Cycle,” Journal of Political Economy 101 (1993),
245–273.

Cascaldi-Garcia, Danilo, “News Shocks and the Slope of the Term Struc-
ture of Interest Rates: Kommentar,” American Economic Review 107
(2017), 3243–3249.

Chang, Andrew C., and Phillip Li, “Measurement Error in Macroeconomic
Data and Economics Research: Data Revisions, Gross Domestic
Product, and Gross Domestic Income,” Economic Inquiry 56 (2018),
1846–1869.

Chari, V. V., Patrick J. Kehoe, and Ellen R. McGrattan, “Are Structural
VARs with Long-Run Restrictions Useful in Developing Business
Cycle Theory?” Journal of Monetary Economics 55 (2008), 1337–
1352.

Chen, Kaiji, and Edouard Wemy, “Investment-Speciﬁc Technology Shocks:
The Source of Long-Run TFP Fluctuations,” European Economic
Rezension 80 (2015), 230–252.

Christiano, Lawrence J., Martin Eichenbaum, and Charles L. Evans, “Nom-
inal Rigidities and the Dynamic Effects of a Shock to Monetary
Policy,” Journal of Political Economy 113 (2005), 1–45.

Christiano, Lawrence J., Martin Eichenbaum, and Robert Vigfusson, “The
Response of Hours to a Technology Shock: Evidence Based on Di-
rect Measures of Technology,” Journal of the European Economic
Association 2 (2004), 381–395.

——— “Assessing Structural VARs,” in Daron Acemoglu, Kenneth Rogoff,
and Michael Woodford, Hrsg., NBER Macroeconomics Annual, 2006,
Bd. 21 (Cambridge, MA: NBER, 2007).

Christiano, Lawrence J., and Terry J. Fitzgerald, “The Band Pass Filter,”

International Economic Review 44 (2003), 435–465.

Cochrane, John H., “Shocks,” Carnegie-Rochester Conference Series on

Public Policy 41 (1994), 295–364.

Comin, Diego, Mark Gertler, and Ana Maria Santacreu, “Technology In-
novation and Diffusion as Sources of Output and Asset Price Fluc-
tuations,” NBER working paper 15029 (2009).

Comin, Diego, and Bart Hobijn, “An Exploration of Technology Diffusion,”

American Economic Review 100 (2010), 2031–2059.

Erceg, Christopher J., Luca Guerrieri, and Christopher Gust, “Can Long-
Run Restrictions Identify Technology Shocks?” Journal of the Eu-
ropean Economic Association 3 (2005), 1237–1278.

Fernald, John, “A Quarterly, Utilization-Adjusted Series on Total Factor
Productivity,” Federal Reserve Bank of San Francisco working paper
series 2012–19 (2014).

Fischer, Jonas, “Comment on: Letting Difference Views about Business Cy-
cles Compete” (S. 457–474), in Daron Acemoglu, Kenneth Rogoff,
and Michael Woodford, Hrsg., NBER Macroeconomics Annual 2009,
Bd. 23 (Cambridge, MA: NBER, 2010).

Francis, Neville, Michael Owyang, Jennifer Roush, and Riccardo DiCe-
cio, “A Flexible Finite-Horizon Alternative to Long-Run Restriction
with an Application to Technology Shocks,” this REVIEW 2014, 96
(4), 638–648.

Gali, Jordi, “Technology, Employment, and the Business Cycle: Do Tech-
nology Shocks Explain Aggregate Fluctuations?” American Eco-
nomic Review 89 (1999), 249–271.

Gort, Michael, and Steven Klepper, “Time Paths in the Diffusion of Product

Innovations,” Economic Journal 92 (1982), 630–653.

Griliches, Zvi, “Hybrid Corn: An Exploration in the Economics of Tech-

nological Changes,” Econometrica 25 (1957), 501–522.

Justiniano, Alejandro, Giorgio Primiceri, and Andrea Tambalotti, “Invest-
ment Shocks and Business Cycles,” Journal of Monetary Economics
57 (2010), 132–145.

Keane, Michael, and Richard Rogerson, “Micro and Macro Labor Supply
Elasticities: A Reassessment of Conventional Wisdom,” Journal of
Economic Literature 50 (2012), 464–476.

Kurmann, André, and Christopher Otrok, “News Shocks and the Slope of
the Term Structure of Interest Rates,” American Economic Review
103 (2013), 2612–2632.

——— “News Shocks and Inﬂation: Lessons for New Keynesians,” Drexel

University technical report (2017A).

——— “News Shocks and the Slope of the Term Structure of Interest Rates:
Reply,” American Economic Review 107 (2017B), 3250–3256.
Kurmann, André, and Elmar Mertens, “Stock Prices, News, and Economic
Fluctuations: Kommentar,” American Economic Review 104 (2014),
1439–1445.

Leeper, Eric M., Todd B. Walker, and Shu-Chun Susan Yang, “Fiscal Fore-
sight and Information Flows,” Econometrica 81 (2013), 1115–1145.
Lindé, Jesper, “The Effects of Permanent Technology Shocks on Hours:
Can the RBC-Model Fit the VAR Evidence?” Zeitschrift für Wirtschaftswissenschaften
Dynamics and Control 33 (2009), 597–613.

Mansﬁeld, Edwin, “Technical Change and the Rate of Imitation,” Econo-

metrica 29 (1961), 741–766.

——— “The Diffusion of Industrial Robots in Japan and the United States,”

Research Policy 18 (1989), 183–192.

Rogers, Everett, Diffusion of Innovations (New York: Free Press, 1995).
Rotemberg, Julio, “Stochastic Technical Progress, Smooth Trends, Und
Nearly Distinct Business Cycles,” American Economic Review 93
(2003), 1543–1559.

Russell, Andrew, and Lee Vinsel, “The Joy of Standards,” New York Times,

Februar 16, 2019.

Smets, Frank, and Raf Wouters, “Shocks and Frictions in US Business
Cycles: A Bayesian DSGE Approach,” American Economic Review
97 (2007), 586–606.

Stock, James H., and Mark W. Watson, “Disentangling the Channels of
the 2007–2009 Recession,” Brookings Papers on Economic Activity
2012 (Frühling), 81–141.

Uhlig, Harald, “What Moves Real GNP?” Humboldt University working

Papier (2003).

Comin, Diego, and Mark Gertler, “Medium-Term Business Cycles,” Amer-

ican Economic Review 96 (2006), 523–551.

——— “Do Technology Shocks Lead to a Fall in Total Hours Worked?”
Journal of the European Economic Association 2 (2004), 361–371.

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
/
R
e
S
T
/