Creative Destruction and Subjective Well-Being Philippe Aghion Ufuk - - PowerPoint PPT Presentation

Creative Destruction and Subjective Well-Being

Philippe Aghion Ufuk Akcigit Harvard UPenn Angus Deaton Alexandra Roulet Princeton Harvard Harvard, April 7th 2014

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 1

Creative Destruction and Subjective Well-Being Introduction

Introduction (1)

Should GDP growth be a primary objective for countries to pursue? The answer is far from being consensual. − → Some argue (e.g. Sen, Stiglitz and Fitoussi (2010)) that indicators other than (per capita) GDP growth should also be taken into account, in particular to reflect environmental quality, unemployment, and income inequality. − → Others take a more radical stand and argue that GDP growth is detrimental to ”happiness”, in particular because it constantly destroys jobs and skills; thus, according to that view, GDP growth should simply be disregarded as a social objective to pursue.

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 2

Creative Destruction and Subjective Well-Being Introduction

Introduction (2)

In this paper, we address this latter objection to growth head on, and investigate whether or when Schumpeterian creative destruction affects subjective well-being positively or negatively.

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 3

Creative Destruction and Subjective Well-Being Introduction

Introduction (3)

To measure creative destruction, we follow Davis, Haltiwanger and Schuh (1996) and use data come from the Business Dynamics Statistics, these data are at the MSA level We consider:

1

Job turnover (job creation rate + job destruction rate)

2

Establishment turnover (highly correlated with job turnover)

3

Creation and destruction rates separately

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 4

Creative Destruction and Subjective Well-Being Introduction

Introduction (4)

The data on subjective well-being come from

1

the Behavioral Risk Factor Surveillance System : life satisfaction question from 2005 onwards.

2

the Gallup Healthways Wellbeing Index : several well-being

• utcomes; starts in 2008.

− → Both have very large sample size: ≈ 350, 000 respondents / year

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 5

Creative Destruction and Subjective Well-Being Introduction

Introduction (5)

To proxy for subjective well-being, we use:

1

the BRFSS life satisfaction question, constructed using the question ”In general how satisfied are you with your life?”

2

the current Cantril ladder of life (from Gallup) , constructed using the question ”imagine a ladder from 0 to 10...on which step do you personally feel you stand at this time?”

3

the anticipated Cantril ladder of life, based the question ”which level of the ladder do you anticipate to achieve in five years?”

We also look at a measure of individuals’ worry

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 6

Creative Destruction and Subjective Well-Being Introduction

Introduction (6)

A higher rate of turnover has both, a direct and an indirect effect

• n life satisfaction

1

Displacement effect: The direct effect is that, everything else equal more turnover should reduce life satisfaction as it translates into a higher probability of becoming unemployed

2

Capitalization effect: The indirect effect is that a higher rate of turnover implies a higher growth externality and therefore a higher net present value of future earnings, which in turn could translate into higher life satisfaction

In this paper we look at how these two effects play out for different types of individuals and also in states with differing unemployment benefit policies

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 7

Creative Destruction and Subjective Well-Being Introduction

Introduction (7): Main findings

Our first finding is that the effect of creative destruction on life satisfaction is unambiguously positive when we control for MSA-level unemployment, less so if we do not Job turnover has an effect on life satisfaction of the opposite sign but of a similar magnitude as that of the unemployment rate It has a substantially larger positive effect on anticipated well-being; yet it is also associated with more ”worry” We find that creative destruction increases life satisfaction more in states with more generous unemployment benefits We also look at interactions with individual characteristics and with the MSA unemployment rate

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 8

Creative Destruction and Subjective Well-Being Introduction

Introduction (7): Related literature

Literature on growth, job turnover and unemployment: e.g see Davis, Haltiwanger, and Schuh (1996), Mortensen and Pissarides (1998), and Aghion and Howitt (1998) − → we contribute to this literature by looking at how turnover affects subjective well-being, not just growth Literature on income and well-being: e.g see Easterlin (1974), Blanchflower and Oswald (2004), Di Tella et al (2007), Deaton (2008), Stevenson and Wolfers (2008), Deaton and Stone (2013) − → we contribute to this literature by putting firms and firm turnover on the RHS of the regression equations

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 9

Creative Destruction and Subjective Well-Being Outline

Outline

1

Introduction

2

Model

3

Data and specification

4

Results

5

Conclusion

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 10

Creative Destruction and Subjective Well-Being Model

Model (1)

Model with Schumpeterian creative destruction which

1

generates growth

2

generates endogenous obsolescence of firms and jobs

Workers in obsolete firms join the unemployment pool until they are matched to a new firm.

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 11

Creative Destruction and Subjective Well-Being Model

Model (2)

Creative destruction has:

1

a positive effect on well-being through economic growth

2

a negative effect on well-being through unemployment due to

• bsolescence

Which of these two effects effect dominates, depends upon individual characteristics (discount rate, degree of risk-aversion,...) and upon local labor market (generosity of unemployment insurance)

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 12

Creative Destruction and Subjective Well-Being Model

Model (3): Production technology

The economy is populated by infinitely-lived and risk-neutral individuals of measure one, and they discount the future at rate ρ = r. The final good is produced according to: ln Yt =

• j∈J ln yjtdj

where J ⊂ [0, 1] is the set of active product lines, with measure J ∈ [0, 1] invariant in steady state Each intermediate firm produces using one unit of labor according to the following linear production function, yjt = Ajtljt, where ljt = 1 is the labor employed by the firm, and the same in all sectors

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 13

Creative Destruction and Subjective Well-Being Model

Model (4): Innovation

An innovator in sector j at date t will move productivity in sector j from Ajt−1 to Ajt = λAjt−1 The innovator is a new entrant, and entry occurs in each sector with Poisson arrival rate x which we take to be exogenous Upon entry in any sector, the previous incumbent firm becomes

• bsolete and its worker loses her job and the entering firm posts a

new vacancy Production in that sector resumes with the new technology when the firm has found a new suitable worker. Thus the measure of inactive product lines is equal to the unemployment rate ut = 1 − Jt, where u denotes the equilibrium unemployment rate.

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 14

Creative Destruction and Subjective Well-Being Model

Model (5): Labor market and job matching

Let m(ut, vt) = uα

t v1−α t

denote the arrival rate of new matches between firms and workers, where ut denotes the number of unemployed at time t and vt denotes the number of vacancies. The flow probability for each unemployed worker to find a suitable firm is m(ut, vt)/ut The flow probability for any new entrant firm to find a suitable new worker is m(ut, vt)/vt Finally, we assume that in each intermediate sector where a worker is currently employed, the worker appropriates fraction β

• f revenues whereas the complementary fraction (1 − β) accrues

to the employer.

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 15

Creative Destruction and Subjective Well-Being Model

Model (6): Valuations and life satisfaction

Our proxy for life satisfaction is the average present value of an individual employee, namely: Wt = utUt + (1 − ut)Et, where:

1

Ut is the net present value of an individual who is currently unemployed

2

Et is the net present value of an individual who is currently employed.

Asset equations: ρEt − ˙ Et = wt + x(Ut − Et) ρUt − ˙ Ut = bt + (m(ut, vt)/ut)(Et − Ut)

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 16

Creative Destruction and Subjective Well-Being Model

Solving the model (1): Equilibrium profits

Logarithmic technology for final good production implies that yjt = Yt/pjt. Then equilibrium profit is simply πjt = pjtyjt − wjt = Yt − wjt where wjt = β

• Yt − wjt
• .

Thus πjt = 1 1 + βYt = πY wjt = β 1 + βYt = βπY

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 17

Creative Destruction and Subjective Well-Being Model

Solving the model (2): Equilibrium unemployment

Our focus is on steady state in which − → all aggregate variables grow at the same constant rate g − → aggregate unemployment u and the number of vacancies remain constant In steady state, the flow out of unemployment must equal the flow into unemployment: m(u, v) = (1 − u)x. In addition, the number of sectors without an employed worker is equal to the number of sectors with an open vacancy, that is: u = v. Therefore: u = x 1 + x

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 18

Creative Destruction and Subjective Well-Being Model

Solving the model (3): Steady state growth

The growth rate of the economy is equal to g = f ln λ, where f denotes the flow of sectors in which a new innovation is being implemented This flow is simply equal to the rate at which new firm-worker matches occur: f = m. Using the fact that in steady-state equilibrium we have: m = u = x 1 + x, we get the equilibrium growth rate: g = x 1 + x ln λ.

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 19

Creative Destruction and Subjective Well-Being Model

Solving the model (4): Equilibrium life satisfaction

Recall that our proxy for life satisfaction is the average present value of an individual employee, namely: W = uU + (1 − u)E, where: rE − ˙ E = βπY + x(U − E) rU − ˙ U = bY + (m(u, v)/u)(E − U)

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 20

Creative Destruction and Subjective Well-Being Model

Solving the model (5): Equilibrium life satisfaction

Now use the fact that in steady state we have:

1

˙ E = gE and ˙ U = gU,

2

m/u = 1

3

u = x/(1 + x)

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 21

Creative Destruction and Subjective Well-Being Model

Solving the model (6): Equilibrium life satisfaction

We then end up with: W = Y r − g

• βπ −

xB 1 + x

• where B ≡ βπ − b.

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 22

Creative Destruction and Subjective Well-Being Model

Solving the model (6): Equilibrium life satisfaction

We then end up with: W = Y r − g

• βπ −

xB 1 + x

• hi

Two effects:

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 22

Creative Destruction and Subjective Well-Being Model

Solving the model (6): Equilibrium life satisfaction

We then end up with: W = Y r − g

• βπ −

xB 1 + x

• hi

1- Capitalization effect: Higher turnover increases the growth rate g which in turns acts positively on life satisfaction.

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 22

Creative Destruction and Subjective Well-Being Model

Solving the model (6): Equilibrium life satisfaction

We then end up with: W = Y r − g

• βπ −

xB 1 + x

• hi

2- Displacement effect: For given growth rate g, more turnover reduces life satisfaction because higher turnover leads to a higher probability of workers losing their current job.

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 22

Creative Destruction and Subjective Well-Being Model

Solving the model (8): First proposition

Proposition (i) A higher turnover rate x always increases life satisfaction W if we abstract from its effect on unemployment; (ii) A higher turnover rate increases life satisfaction more the lower the discount rate ρ, i.e: ∂2W ∂x∂ρ < 0 and life satisfaction increases with turnover when ρ < βπ ln λ

B

, and it decreases with turnover otherwise; (iii) Life satisfaction increases more with turnover when unemployment benefits are more generous. i.e: ∂2W ∂x∂b > 0.

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 23

Creative Destruction and Subjective Well-Being Model

Solving the model (9): Risk aversion

We now consider risk averse individuals with U = ln C. Now well-being can be shown to be equal to: Wu(c)=ln c = 1 ρ

• x

1 + x ln (b) + 1 1 + x ln (βπ)

• + 1

ρ g ρ + ln Y

• Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard)

24

Creative Destruction and Subjective Well-Being Model

Solving the model (10): Second proposition

Proposition A higher turnover rate has a less positive effect on agents that are risk averse with U = ln C than on risk-neutral agents: ∂Wu(c)=ln c ∂x < ∂Wu(c)=c ∂x . Moving continuously from the baseline case where individuals are risk-neutral towards the risk-averse case where individuals have log preferences, makes the effect of creative destruction on life satisfaction become increasingly less positive (or increasingly more negative)

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 25

Creative Destruction and Subjective Well-Being Model

Extensions

Endogenous entry Exogenous job destruction Transitions

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 26

Creative Destruction and Subjective Well-Being Data

Data (1)

The data on job turnover and creative destruction − → come from the Business Dynamics Statistics, which provides, at the metropolitan (MSA) level, information on job creation and destruction rates as well as on the entry and exit rates of establishments − → these rates are computed from the whole universe of firms as described in the Census Longitudinal Business Database

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 27

Creative Destruction and Subjective Well-Being Data

Data (2)

From that database we look at:

1

Job turnover rate (sum of job creation rate and job destruction rate)

2

Establishment turnover rate (sum of the establishment entry rate and the establishment exit rate) − → these two measures are highly correlated and yield very similar results

3

Creation and destruction rates separately

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 28

Creative Destruction and Subjective Well-Being Data

Data (3)

The data on subjective well-being comes from

1

the Gallup Healthways Wellbeing Index survey, which asks each day several distinct questions on subjective well-being to 1,000 randomly selected individuals . It starts in 2008.

2

the Behavioral Risk Factor and Surveillance System started asking a ”life satisfaction” question in 2005. Both have very large sample size: ≈ 350, 000 respondents / year

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 29

Creative Destruction and Subjective Well-Being Data

Data (4)

To proxy for subjective well-being in Gallup-Healthways, we use

1

the current Cantril ladder constructed based on the question ”Imagine a ladder from 0 to 10[...]on which step of the ladder would you say you personally feel you stand at this time?”

2

the anticipated Cantril ladder based on the question ”What level

• f the ladder do you anticipate to achieve in five years?”

3

We also investigate how creative destruction affects a measure of individuals ”worry”, based on binary answers to the question ”Did you experience worry during a lot of the day yesterday?”

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 30

Creative Destruction and Subjective Well-Being Data

Data (5)

To proxy for subjective well-being in the BRFSS, we use the Life satisfaction question : ”In general how satisfied are you with your life?” The possible answers are ”Very satisfied” (1), ”Satisfied” (2) ”Disatisfied” (3), ”Very disatisfied” (4) We recoded them so that an increase in the variable means an increase in subjective well-being

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 31

Creative Destruction and Subjective Well-Being Data

Summary statistics - subjective well-being

Mean Standard deviation Min Max Current ladder (Gallup) 6.78 1.95 10 MSA-level averages 6.78 0.14 6.15 7.51 Anticipated ladder (Gallup) 8.05 1.99 10 MSA-level averages 8.05 0.15 7.42 8.48 Worry (Gallup) 0.32 0.47 1 MSA-level averages 0.32 0.02 0.22 0.40 Life satisfaction (BRFSS) 3.37 0.63 1 4 MSA-level averages 3.37 0.047 3.14 3.58

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 32

Creative Destruction and Subjective Well-Being Data

Summary statistics -creative destruction

(2005-2010 averages) Mean Standard deviation Min Max Job turnover 0.29 0.035 0.18 0.43 Job creation rate 0.15 0.015 0.08 0.22 Job destruction rate 0.14 0.017 0.09 0.22 Unemployment rate 0.065 0.015 0.03 0.24

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 33

Creative Destruction and Subjective Well-Being Data

.1 .12 .14 .16 .18 .2 Job creation rate .1 .12 .14 .16 .18 .2 Job destruction rate

2005-2010 averages

Job creation and Job destruction

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 34

Creative Destruction and Subjective Well-Being Data

.05 .06 .07 .08 .09 .1 Unemployment rate .2 .25 .3 .35 .4 Job turnover (creation + destruction rates)

2005-2010 averages

Unemployment and Job turnover

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 35

Creative Destruction and Subjective Well-Being Empirical framework

Empirical framework

1

MSA-level regressions of creative destruction on subjective well-being − → Across years averages to mirror the steady-state analysis of the model

2

Individual level regressions − → Rich set of controls for individual determinants of well-being

3

Robustness checks

4

Interactions with individual characteristics and with state-level policy (unemployment insurance)

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 36

Creative Destruction and Subjective Well-Being Empirical results

Metropolitan Statistical Area (MSA) Results 1/3

Table 2 - MSA-level averages; BRFSS data - 2005-2010 averages (1) (2) (3) (4) (5) VARIABLES ”How satisfied are you with your life?” Unemployment Rate

• 1.566***
• 1.731***
• 1.608***

(0.198) (0.199) (0.216) Job turnover 0.0970 0.279*** (0.0971) (0.0723) Job creation rate 1.360*** 0.688*** (0.239) (0.214) Job destruction rate

• 1.424***
• 0.242

(0.282) (0.270) Observations 323 323 323 323 323 R-squared 0.305 0.007 0.356 0.110 0.366

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 37

Creative Destruction and Subjective Well-Being Empirical results

Metropolitan Statistical Area (MSA) Results 2/3

Table 3 - MSA-level averages; Gallup data (2008-2011) - (1) (2) (3) (4) (5) VARIABLES Current ladder Unemployment Rate

• 2.678***
• 3.428***
• 2.421***

(0.566) (0.580) (0.550) Job turnover 0.526 1.303*** (0.368) (0.370) Job creation rate 6.454*** 4.809*** (1.107) (0.981) Job destruction rate

• 4.482***
• 2.080***

(0.774) (0.700) Observations 363 363 363 363 363 R-squared 0.139 0.014 0.212 0.190 0.265

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 38

Creative Destruction and Subjective Well-Being Empirical results

Metropolitan Statistical Area (MSA) Results 3/3

Table 3 - MSA-level averages; Gallup data (2008-2011) - (1) (2) (3) (4) (5) VARIABLES Anticipated ladder Unemployment Rate

• 0.499
• 1.872***
• 1.274**

(0.529) (0.469) (0.513) Job turnover 1.961*** 2.385*** (0.291) (0.319) Job creation rate 5.332*** 4.467*** (0.896) (0.884) Job destruction rate

• 0.887

0.377 (0.741) (0.875) Observations 363 363 363 363 363 R-squared 0.004 0.167 0.220 0.218 0.236

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 39

Creative Destruction and Subjective Well-Being Empirical results

Magnitude - MSA-level regressions

Going from an MSA at the 25th percentile in terms of job turnover (i.e. with a job creation rate + job destruction rate at 26.8 % ) to an MSA at the 75th percentile (i.e with a job turnover at 31.8% ) is associated with an increase in life satisfaction of 0.9 % with respect to the mean and with an increase in anticipated well-being of 1.5% As a benchmark going from an MSA at the 25th percentile in terms of unemployment rate to an MSA at the 75th percentile is associated with a decrease in current satisfaction of 0.8 % and a decrease in anticipated well-being of 0.3 %

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 40

Creative Destruction and Subjective Well-Being Empirical results

Individual Level Regressions

The specification at the individual level is: SWBi,m,t = δCDm,t + αUm,t + βXi,m,t + Tt + ǫi,s,t, Individual controls include : gender, ethnicity, detailed education and family status, age, age2 Year and Month Fixed effect Standard errors clustered at the MSA level We restrict attention to working-age individuals (18-60 years old)

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 41

Creative Destruction and Subjective Well-Being Empirical results

Individual Level 1/4 - Life satisfaction (BRFSS)

(1) (2) (3) (4) (5) VARIABLES ”How satisfied are you with your life?” Unemployment Rate

• 0.682***
• 0.757***
• 0.737***

(0.105) (0.106) (0.112) Job turnover 0.139** 0.192*** (0.0603) (0.0564) Job creation rate 0.293*** 0.234*** (0.0865) (0.0812) Job destruction rate

• 0.0276

0.144* (0.0757) (0.0733) Month F.E. x x x x x Year F.E. x x x x x Observations 592,501 592,452 592,452 592,452 592,452 R-squared 0.077 0.076 0.077 0.077 0.077

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 42

Creative Destruction and Subjective Well-Being Empirical results

Individual level results 2/4 - Current ladder (Gallup)

(1) (2) (3) (4) (5) VARIABLES ”Current ladder” Unemployment Rate

• 2.446***
• 2.878***
• 2.530***

(0.421) (0.431) (0.422) Job turnover 0.254 0.752*** (0.246) (0.230) Job creation rate 1.561*** 1.044*** (0.440) (0.351) Job destruction rate

• 0.765***

0.211 (0.289) (0.268) Month F.E. x x x x x Year F.E. x x x x x Observations 502,334 502,334 502,334 502,334 502,334 R-squared 0.058 0.058 0.058 0.058 0.058

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 43

Creative Destruction and Subjective Well-Being Empirical results

Individual level results 3/4 - Anticipated ladder

(1) (2) (3) (4) (5) VARIABLES ”Anticipated ladder” Unemployment Rate 0.108

• 0.705**
• 0.677**

(0.357) (0.307) (0.307) Job turnover 1.319*** 1.441*** (0.154) (0.151) Job creation rate 1.601*** 1.516*** (0.275) (0.259) Job destruction rate 1.099*** 1.373*** (0.230) (0.218) Month F.E. x x x x x Year F.E. x x x x x Observations 490,316 490,086 490,086 490,086 490,086 R-squared 0.077 0.077 0.077 0.077 0.077

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 44

Creative Destruction and Subjective Well-Being Empirical results

Individual level results 4/4 - Worry

(1) (2) (3) (4) (5) VARIABLES ”Worry” Unemployment Rate 0.417*** 0.367*** 0.357*** (0.0712) (0.0759) (0.0784) Job turnover 0.159*** 0.0952** (0.0419) (0.0408) Job creation rate 0.0245 0.0689 (0.0747) (0.0672) Job destruction rate 0.263*** 0.119** (0.0554) (0.0569) Month F.E. x x x x x Year F.E. x x x x x Observations 503,395 503,159 503,159 503,159 503,159 R-squared 0.014 0.013 0.014 0.014 0.014

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 45

Creative Destruction and Subjective Well-Being Empirical results

Magnitude - individual-level regressions

Going from an MSA at the 25th percentile in terms of job turnover (i.e. with a job creation rate + job destruction rate at 26.8 % ) to an MSA at the 75th percentile (i.e with a job turnover at 31.8% ) is associated with an increase in life satisfaction of 0.5 % with respect to its mean and with an increase in anticipated well-being of 0.9% As a benchmark going from an MSA at the 25th percentile in terms of unemployment rate to an MSA at the 75th percentile is associated with a decrease in life satisfaction of 0.6 % and a decrease in anticipated well-being of 0.1 %

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 46

Creative Destruction and Subjective Well-Being Empirical results

Robustness checks 1/3

1

We control for the sectoral composition of the MSA

2

We look at establishment turnover

3

We use a Bartik-type measure of creative destruction

• CDm,t = ∑

i

ωi,m,2004 × CDi,USA,t

ωi,m,2004 is derived from the sectoral distribution of employment in MSA m in 2004 CDi,USA,t are the nationwide measures of creative destruction for each sector of activity

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 47

Creative Destruction and Subjective Well-Being Empirical results

Robustness checks 2/3

The specification is: SWBi,m,t = δCDm,t + αUm,t + βXi,m,t + Tt + ∑

i

θωi,m,2004 + ǫi,s,t, where CD is either Job turnover or Establishment turnover or the predicted value of either of them CDm,t = ∑i ωi,m,2004 × CDi,USA,t Controls include previous individual controls ( gender, ethnicity, detailed education and family status, age, age2, income), year and Month Fixed effect, the initial sectoral composition Standard errors clustered at the MSA level

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 48

Creative Destruction and Subjective Well-Being Empirical results

Robustness checks 3/3

(1) (2) (3) (4) VARIABLES ”How satisfied are you with your life?” Job turnover 0.126** (0.0633) Estab turnover 0.244** (0.0952) Predicted job turnover 0.932*** (0.313) Predicted estab turnover 1.875*** (0.610) Unemployment Rate

• 0.732***
• 0.727***
• 0.669***
• 0.684***

(0.122) (0.122) (0.121) (0.121) Individual controls x x x x Initial Sectoral composition x x x x Year and Month F.E. x x x x Observations 925,556 925,556 925,150 925,150 R-squared 0.001 0.002 0.097 0.097 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 49

Creative Destruction and Subjective Well-Being Empirical results

Interactions

At the individual level: with the employment status + with other characteristics At the state level: with the unemployment insurance

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 50

Creative Destruction and Subjective Well-Being Empirical results

Interaction with individual employment status 1/2

The specification is: SWBi,m,t = δCDm,t + γCDm,t ∗ Empi,m,t + θEmpi,m,t + αUm,t + βXi,m,t + Tt + ǫi,s,t, ”Employed” are wage earners; ”Not employed” are either unemployed or out of the labor force; Recall that we restrict ourselves to working age people (18-60 years old) so retirees not included Self-employed are dropped from the analysis but results are similar if we include them in the employed group Individual controls include : gender, ethnicity, detailed education and family status, age, age2 Year and Month Fixed effect Standard errors clustered at the MSA level

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 51

Creative Destruction and Subjective Well-Being Empirical results

Interaction with individual employment status 2/2

(1) (2) (3) (4) (5) (6) VARIABLES ”How satisfied are you with your life?” Creation rate

• 0.617***
• 0.252*
• 0.619***
• 0.252*

∗ employed (0.149) (0.133) (0.147) (0.132) Destruction rate 0.130 0.190 0.0913 0.170 ∗ employed (0.167) (0.154) (0.169) (0.156) Job turnover

• 0.316***
• 0.0724

∗ employed (0.117) (0.109) Job creation rate 0.745*** 0.467*** 0.674*** 0.417*** (0.180) (0.161) (0.167) (0.156) Job destruction rate

• 0.190
• 0.123

0.105 0.0365 (0.142) (0.122) (0.138) (0.128) Job turnover 0.429*** 0.251** (0.115) (0.110) (0.132) (0.108) (0.141) (0.116) Unemployment rate x x x x

• Indiv. controls

x x x Month and Year F.E. x x x x x x Observations 536,387 536,341 536,387 536,341 536,387 536,341 R-squared 0.020 0.087 0.019 0.087 0.020 0.087

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 52

Creative Destruction and Subjective Well-Being Empirical results

Interaction with unemployment insurance (1/3)

Next, we split our sample between states with higher than median generosity in unemployment benefits and states with lower than median generosity in unemployment benefits We define generosity as the maximum weekly unemployment benefit amount

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 53

Creative Destruction and Subjective Well-Being Empirical results

Unemployment Insurance 2/3

• .1
• .05

.05 .1 Life satisfaction, conditional on the unemployment rate .2 .25 .3 .35 .4 Job creation + Job destruction rates

A dot represents the mean of the y variable for each equal-sized bins in terms of the x variable The generosity of UI is proxied by the maximum weekly benefit amount

MSAs in states below median in terms of Unemployment Insurance

Life satisfaction and Job Turnover

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 54

Creative Destruction and Subjective Well-Being Empirical results

Unemployment Insurance 3/3

• .1
• .05

.05 .1 Life satisfaction, conditional on the unemployment rate .2 .25 .3 .35 .4 Job creation + Job destruction rates

A dot represents the mean of the y variable for each equal-sized bins in terms of the x variable The generosity of UI is proxied by the maximum weekly benefit amount

MSAs in states above median in terms of Unemployment Insurance

Life satisfaction and Job Turnover

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 55

Creative Destruction and Subjective Well-Being Empirical results

Interaction with other individual-level characteristics

We then perform regressions where we split our sample:

1

between the religious and the non-religious

2

between the smokers and the non-smokers

The specification is: SWBi,m,t = δCDm,t + γCDm,t ∗ Zi,m,t + θZi,m,t + αUm,t + βXi,m,t + Tt + ǫi,s,t, Z is a dummy for either religiosity or smoking Individual controls include : gender, ethnicity, education and family status, age, age2 and log of income Year and Month Fixed effect Standard errors clustered at the MSA level

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 56

Creative Destruction and Subjective Well-Being Empirical results

Religion

(1) (2) (3) (4) VARIABLES Anticipated ladder Religious * Job turnover

• 0.506**
• 0.513**

(0.207) (0.206) Religious * Creation rate 0.433 0.424 (0.350) (0.349) Religious * Destruction rate

• 1.164***
• 1.167***

(0.255) (0.255) Job turnover 1.341*** 1.421*** (0.201) (0.201) Job creation rate 0.760** 0.715** (0.323) (0.322) Job destruction rate 1.743*** 1.931*** (0.277) (0.279) Religious 0.276*** 0.278*** 0.251*** 0.252*** (0.0578) (0.0575) (0.0585) (0.0583) Unemployment Rate x x Year F.E. x x x x Month F.E. x x x x Observations 614,296 614,296 614,296 614,296

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 57

Creative Destruction and Subjective Well-Being Empirical results

Smoking

(1) (2) (3) (4) VARIABLES Anticipated ladder Smoker *Job turnover 0.191 0.188 (0.292) (0.292) Smoker * Creation rate

• 0.480
• 0.490

(0.469) (0.468) Smoker * Destruction rate 0.674* 0.677* (0.359) (0.358) Job CD 0.945*** 1.028*** (0.118) (0.123) Job creation 1.114*** 1.060*** (0.213) (0.205) Job destruction 0.823*** 1.027*** (0.178) (0.186) Smoker

• 0.202**
• 0.201**
• 0.185**
• 0.184**

(0.0811) (0.0810) (0.0818) (0.0817) Unemployment Rate x x Year F.E. x x x x Month F.E. x x x x Observations 616,069 616,069 616,069 616,069

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 58

Creative Destruction and Subjective Well-Being Conclusion

Conclusion (1): Summary

We have analyzed the relationship between turnover-driven growth and subjective well-being, using MSA level turnover data from the Longitudinal Business Database and subjective well-being data from Gallup-Healthways and from the BRFSS We found that the overall effect of turnover (creative destruction)

• n subjective well-being is unambiguously positive when we

control for MSA-level unemployment, less so if we do not We found that creative destruction has an effect of opposite sign but roughly similar magnitude as that of the unemployment rate and has a stronger postiive effect on anticipated well-being We found that creative destruction has a more positive effect on subjective well-being in states with more generous unemployment benefits We looked at how creative destruction interacts with individual characteristics

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 59

Creative Destruction and Subjective Well-Being Conclusion

Conclusion (2): Extensions

1

Compare more systematically the determinants of (per capita) GDP growth with the determinants of life satisfaction

2

Look at other individual characteristics or characteristics of labor market (training systems, availability of vocational education,..) which should also impact on the effects of turnover on subjective well-being

3

Gather more data in order to perform event studies − → e.g track a same individual through successive periods of employment and unemployment and look how the well-being indicators for that individual evolve over time as this individual moves back and forth between employment and unemployment

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 60

Creative Destruction and Subjective Well-Being Back up

Non crisis years: 2005-2007

Table 2 - MSA-level averages; BRFSS data - (1) (2) (3) (4) (5) VARIABLES ”How satisfied are you with your life?” Unemployment Rate

• 1.976***
• 2.004***
• 1.819***

(0.283) (0.274) (0.268) Job turnover 0.134 0.166** (0.0985) (0.0753) Job creation rate 0.816*** 0.620*** (0.172) (0.143) Job destruction rate

• 1.061***
• 0.637***

(0.254) (0.211) Observations 317 317 317 317 317 R-squared 0.241 0.012 0.259 0.106 0.299

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 61

Creative Destruction and Subjective Well-Being Back up

Crisis years : 2008-2010

Table 2 - MSA-level averages; BRFSS data - (1) (2) (3) (4) (5) VARIABLES ”How satisfied are you with your life?” Unemployment Rate

• 1.154***
• 1.348***
• 1.457***

(0.151) (0.173) (0.193) Job turnover 0.0656 0.329*** (0.115) (0.110) Job creation rate 0.684** 0.0124 (0.289) (0.257) Job destruction rate

• 0.417*

0.614** (0.236) (0.238) Observations 318 318 318 318 318 R-squared 0.220 0.002 0.268 0.030 0.275

Aghion (Harvard), Akcigit (UPenn), Deaton (Princeton), and Roulet (Harvard) 62