You are on page 1of 40

The Rise of Alternati

Alternative
e Work
Arrangements & the Gig Economy
Lawrence F. Katz
Harvard University
(based on joint work with Alan
Krueger)
March 14, 2016

Overview: Trends
Online Gig Economy has been growing very rapidly
Uber is the q
quintessential employer
p y of g
gig
g work & could
represent to 2/3rds of all gig work
But Offline alternative work swamps Online Gig Work
Indirect impacts of IT and big data applied to HR is
growth of offline alternative work (contracting out,
i d
independent
d t contractors,
t t
oncall
ll workers)
k )
Offline alternative work accelerated last decade around
a 50% rise over last 10 years
years, with esp
esp. fast growth in
contracting out (Katz and Krueger 2016)
All net U.S. employment growth since 2005 appears to be
in alternative work arrangements

Overview: Potential Implications

Alternative work may provide more


flexible work arrangements for many to
enhance work-life balance
But real worries as well: Growth in
alternative work arrangements can put
pressure on wages
g and labor
downward p
standards
How extend social compact
p
between
workers and companies to nonstandard
employment settings?

Our Conventional Labor Force Data are Not Well


Designed to Capture Alternative Work Arrangements
CPS misses growth in self employment compared
with
i h IRS d
data (S
(Schedule
h d l C fil
filers no employees
l
)
1099 filings also on the rise
Multiple job holding still small, but CPS misses
many workers who do occasional gigs or small
tasks online
These problems will become much more
significant with growth of sharing
sharing economy
economy
Need for more frequent and comprehensive
Contingent and Alternative Work Survey

Disturbing Divergence: IRS and CPS data on SelfE l


Employment
as a Sh
Share off CPS T
Totall E
Employment
l
18%
16%

ScheduleC
Filings

14%
12%
10%

CPSUnincorp.
SelfEmployed

8%
6%
4%
2%
0%
1979

1982

1985

1988

1991

1994

1997

2000

2003

2006

2009

2012 2014

Self-Employed Non-Employers (with $1000 or more in


Schedule C income) as Share of Employment, 1997-2013

Source:Abrahametal.(2015)

Filers With 1099-Misc Income As a Share of CPS Total


Employment Show Modest Rise
14%

13%

12%

11%

10%

9%
%

8%
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

Multiple Job Holding as a Percent of All


Employment; CPS 1994-2016

Source:BureauofLaborStatistics.

Katz & Krueger Survey of 2,300 Workers


on Amazons Mechanical Turk
Summer 2015
Asked standard CPS questions on employment and
multiple
p jjob holding
g and additional q
question: Did
you work on any gigs, HITs or other small paid
jobs last week that you did not include in your
response to the previous question?

Many Small Jobs in Sharing Economy are


Missed by Traditional Labor Force Questions

Source:PreliminarytabulationofKatzandKrueger2015survey.Note:1,397
respondentsreportedholdingonlyonejoblastweek,and894reportedholding
multiplejobslastweek.

Katz and Krueger Rand American Life Panel Survey


Conducted October/November 2015
N=3,844; Internet panel.
CPS Contingent & Alternative Work Survey Questions (last done by
BLS in
i 2005)) + Questions
Q ti
on W
Work
k iin Gig
Gi E
Economy
Findings Regarding Online-Intermediated Work:
19.4%
9
of workers do direct selling
g on their main or secondaryy jjob
Only 7% of direct sellers work through an intermediary like Uber
or Avon
One third of those who sell through an intermediary use an online
app, like Uber or Taskrabbit
So, only 0.5% of all workers in online gig economy in reference week
About 1.5% of workers acknowledged omitting gigs or small jobs in
previous week, but most of these were not conducted through an
online intermediary,
intermediary like Uber or Taskrabbit.
Taskrabbit Under reporting raises
multiple job holding by about 10%.

Share of Workers Participating in Online Platform Labor


Market is Still Small (<1%) But Growing Fast

Source: Diana Farrell and Fiona Greig, JPMorgan Chase & Co. Institute, Paychecks, Paydays and the Online
Pl tf
Platform
Economy,
E
Feb.
F b 2016.
2016
Note: Based on monthly deposits of 30 large online work platforms.

Estimating the Size of the Online Gig Economy:


Benchmarking to Uber and Google Searches

Source:JonathanHallandAlanKrueger(2015)andJuddCramer(2015).

Google Trends: 4-Week Moving Average of


Web Searches
Index (Uber)
90
Uber (L eft Scal e)
80
70
60
50

Index (Relative to Uber)


18

"axiom law"
ChaCha
gocurb+"t axi m agi c"
fi verr
Grub Hub
hourlynerd+"hourly nerd"
Lyft
medi cast
S amas ource
S idecar
TaskRabb it
up work+el an ce+odesk

agentany thi ng
tryc aviar+"c avi ar deliv ery"
c lickwork er
eden m cc allum
Gengo
Handy
Inst acart
m echani cal tu rk+mtur k+"am az on tu rk"
redbeacon+"red
redbeacon+
red beacon"
beacon - light
S hyp
s ki llshare
Th um btack
washi o

16
14
12
10

40

30

6
Week
Ended
11/7/15

20
10
0
Jan-04

4
2

Jan-06

Jan-08

Jan-10

Jan-12

Jan-14

0
Jan-16

Google Trends: Cumulative Web Searches in 2015


Index (Relative to Uber)
100

100.0

90

SearchesforUberare2XEverythingElse
C bi d (100 48 5)
Combined(100v.48.5)

80
70

IfscalerelativetoUbers~400,000drivers,
only ~600
only
600,000workersinwholesector,or
000 workers in whole sector or
0.4%oftotalU.S.employment.

60
50
40
30
18.5

20

12.4

10

Was hio
W

S kil lshare

S hyp

S amas ourc e

Red B e
eacon

Med
di cas t

Instt acart

Hourlyy Nerd

Gengo
G

E den McC
Callum

Curb

Clickw
worker

Cavi ar
C

A xiom

A gent A nyy thing

Ch
haCha

Handy
H

S idecar

TaskR
Rabbi t

Mecha
anical
Tu rk

Thum btack

Fiverr

Up
pwork

Lyft

Gru
ubHub

Uber

3.5 3.2 3.1 2.9 1.7


1.4 1.1 0.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Katz and Krueger Rand American Life Panel Survey


Alternative Work

Rand developed weights to match CPS totals by age, education, race,


gender & income.

Reweighted for oversample of self-employed

CPS Contingent & Alternative


l
Work
k Supplement
l
(CWS)
(
) Questions
Conducted 1995, 1997, 1999, 2001, 2005

Attempted to make CWS as comparable to Rand data as possible (e.g.,


imposed same skip logic and drop day laborer).

In CWS, only asked Independent Contractor/Freelancer (IC) question if


say no to Temp Help, On-Call and Contracted Out. In CWS and Rand,
cant be On-Call and Contracted Out. Rand asks IC of everyone.

RAND Survey Key Questions on Alternative Work


Arrangements

On-Call Work: Were you an ON-CALL worker last week? On-call


means you have certain days or hours during which you are not at
work
k but
b on standby
db until
il called
ll d to work.
k

Contracted-Out Work: Did you work for a company that


contracted out you or your services last week?

Independent
p
Contractors: Last week,, were you
y working
g or
self-employed as an independent contractor, an independent
consultant, or a freelance worker? That is, someone who obtains
customers on their own to provide a product or service.

Temporary Help Agency Employees: Are you paid by a


temporary
p
y help
p agency?
g
y

Table1:ComparingRandSampletoCPS ReweightedSampleLooksReasonablyRepresentative

Growth in Alternative Work Arrangements Accelerated


Last Decade , Especially Contracting Out Workers
Alternative Employment Arrangements
Percent of Total Employed Who Worked During Survey Week
20
18
16
14

Workers Provided by Contract Firms


Temporary Help Agency Workers
On-Call Workers (excl. Day Laborers)
Independent Contractors
Total

12
10
8

10.1

15.8
31
3.1
1.6

9.3
0.5
1.0
1.6

0.6
09
0.9

2.6

1.7

6
4

8.4
6.4

69
6.9

1995

2005

2
0
2015 Estimate
(Alt. Weight)

Note: Individuals can be categorized as both on-call workers and workers provided by contract firms. Alternative weights were
estimated for 2015 data to match the share of self-employed workers in the October 2015 Current Population Survey.
Source: Bureau of Labor Statistics; 2015 Katz and Krueger Rand survey.

Alternative Work Arrangements: 1995 to 2015

U.S. Employment Levels and Structure: 2005 & 2015


EmploymentType
Standard

Alternative

Total

Feb2005

126.2

14.2

140.4

Oct2015

125.8

23.6

149.4

% 2005 15
%200515

0.3%
0 3%

+66.5%
66 5%

+6.5%
6 5%

Employmentinmillions
Slightlyover100%ofU.S.netemploymentgrowthsince2005innonstandard
work
Sources:BLSLaborForceStatisticsfortheCPS,2005CWS,2015KatzKrueger
RANDSurvey

Table3:CharacteristicsofWorkersinAlternativeWorkArrangements

Where Has Incidence of Alternative Work Grown the


Most? (See Table 4)
Micronumerosity (small n) Warning
Old workers
Older
k
((esp. ffreelancers)
l
)
Women
Whites Blacks & Hispanics
Whites,
Occ/Industry: Gains in manufacturing, retail,
health care support
support, education
education, social services,
services
transportation, computers
Contracting
g out education, health and p
public
administration followed professional business services,
janitorial and construction in contracting out more

Percent in Alternative Work Arrangements, by Age


30
23.9

Percent

25
20
14.3

15

13.8 14.4

9.3 9.8

10
5
0

2554

2005
2015

5.9 6.4 6.4

1624

1995

5575

A C
AgeCategory
Source: Bureau of Labor Statistics; 2015 Katz and Krueger Rand survey.

Percen
nt

Percent in Alternative Work Arrangements,


by Sex, Race/Ethnicity & Education
20

20

15

15
1995

10

2005

2015

10
5
0

Men

Women

White

Black

20
15
1995

10

2005

2015

0
<HS

HS

Some College+
College

Source: Bureau of Labor Statistics; 2015 Katz and Krueger Rand survey.

Hispanic

Percent of Workers in Alternative Work Arrangements,


By Industry
30.0

25 0
25.0
1995
20.0
2005
2015

15.0

10.0

5.0

0.0

Source: Bureau of Labor Statistics; 2015 Katz and Krueger Rand survey.

Percent of Workers in Alternative Work Arrangements,


By Occupation
40.0

35.0

1995

30.0

2005
25.0

2015

20.0

15.0

10.0

5.0

0.0

Source: Bureau of Labor Statistics; 2015 Katz and Krueger Rand survey.

Summary Measures Based on Predicted Wages


Estimate a kitchen sink Mincer wage regression with
2005 CPS ORG file (Xs included education, experience,
race sex
race,
sex, industry
industry, and occupation)
Determine quintile cutoffs of y-hat based on 2005 CPS
Predict y-hat for every observation in 1995, 2005, 2015
CWS and
dR
Rand
dS
Survey
Compute probability of being in alternative work by
quintile of y-hat
Note
N
ffurther
h adjustment
dj
to weights
i h so quintiles
i il off y-hat
h
in Rand sample as a whole match October 2015 CPS
shares as a whole. Rand under sampled low quintiles.
Summarize the part of the (predicted) wage distribution
where alternative work has grown, and the relationship
bet een alternative
between
alternati e work
ork and predicted wages
ages

Percent in Alternative Work Arrangement


by Predicted Earnings Quintile

Percent in Temporary Help Jobs


by Predicted Wage Quintile

Percent On-Call Workers


by Predicted Wage Quintile

Percent Contracted Out


by Predicted Wage Quintile

Percent Independent Contractor/Freelancers


by Predicted Wage Quintile

Wages and Hours of Nonstandard Workers in 2015


Run augmented Mincerian human capital wage
regressions to examine hourly and weekly wage
premiums for workers in alternative work
arrangements -- control for education, experience,
occupation demographics
occupation,
Independent contractors earn a 17% hourly wage
premium but have 28% lower weeklyy earnings
p
g in main
job from lower hours than standard workers
Contracted out workers earn similar hourly wages to
standard
t d d workers
k
b
butt 26%
6% llower weekly
kl earnings
i
Workers in alternative arrangements more than twice
as likely to be involuntary part
part-time
time (7
(7.7%
7% vs
vs. 3
3.3%)
3%)

Preliminary Exploration of
Reasons for the Rise of Alternative Work
Demographic shifts (~10%)
Misclassification to avoid providing benefits and
legal protections, and/or to reduce rent sharing
(David Weils fissuring;
g; Abraham and Taylor
y
(1996))
Disintermediation due to technological
changes/scale
h
/ l economies
i
Fallout from Great Recession
Smoothing of employment to reduce volatility for
core employees in face of demand shocks
((Abraham and Taylor).
y )

Alternative Work Shift Share Analysis

Almost Half of Increase is Due to the Rise in the


Contracting Out Component
Rose 2.5 pp -- from 0.6% of workers to 3.1% from
2005 to 2015.
2015
Contract work was the only category of alternative
work that saw a meaningful rise among younger
workers
occupat o s with
t tthee largest
a gest increase
c ease in tthee
Thee 5 occupations
incidence of contract workers (computers;
community & social service; arts, design &
entertainment; protective security; construction)
accounted for 36 percent of the overall rise.

Possible Policy Implications I


Labor market is evolving in ways not fully captured
in official statistics, h employment in the gig
economy is still very small but rapidly growing.
growing
Does contracting out and increasing use of
independent contractors threaten labor standards
and contribute to rising wage inequality?
Schmieder (2015) for Germany; Dube-Kaplan (2010)
for U.S.
Public and Private policy responses? Third-party
activism as well
Harvard Wage and Benefits Parity Policy (WBPP) as
a potential model to allow contracting out for
efficiency gains but not to undercut labor standards

Public Policy Implications II


Worker mislcassification by employers should be
vigorously policed through regulatory
enforcement, but gray area presents complex set of
issues.
Independent Contractor/Employee Dichotomy
becoming outdated, and creates uncertainty for
companies and workers.
Legitimate Gray area: e.g., Uber or Avalon
Harris-Krueger Proposal: Create a new category of
Independent Workers that covers gray area
between employee and independent contractor -goal to maintain social compact and efficiency.

Conclusions
Rising inequality and job polarization occurred in the
old
old economy
economy in U.S. before rapid growth in
alternative work (at least based on CWS).
The number of jobs in the online gig economy growing
very fast from a low base.
In the last decade, however, there has been greater
proliferation
lif ti off offline
ffli th
than online
li alternative
lt
ti work
k
arrangements.
Careful thought needs to go into reforming labor laws
to support innovation and maintain the social
compact. This applies to Independent Contractors
working through OnLine and OffLine intermediaries.

You might also like