Temperature  Shocks  and  Firm  Dynamics  in  Developing  Countries  :  Evidence  from  Cote  d’Ivoire  

 Results  are  preliminary,  please  do  not  cite.  

 Nouhoum  Traore    

   University  of  Wisconsin  Madison    

Paris  School  of  Economics            

1    

Outline  

q Research  QuesIon  q MoIvaIon  q Evidence  of  Warming  q Literature  Review  q Model  q Empirical  Strategy  q Data  q Preliminary  Results  q Conclusion  q Work  in  Progress  

 

Research  QuesIons  

q  What  is  the  effect  of  increased  temperatures  on  firm’s  producIvity  in  Cote  d’Ivoire?      

 q  If  any,  what  are  the  potenIal  channels  of  transmission  of  such  effect?    q  Does  it  affect  firm’s  survival?    q What  is  its  implicaIon  for  firm’s  exportability?  

   

 

     

MoIvaIon  

q  There  are  clear  evidences  of  global  warming.  The  planet  is  0.8  ̊C  warmer  today  than  in  pre-­‐industrial  Imes  and  it  could  be  2  ̊C  warmer  in  20  to  30  years  from  now  (World  Bank,  2013).  How  this  would  affect  poor  countries?    

 q  Growing  evidence  of  the  impact  of  temperature  on  the  manufacturing  

sector  (Sunarshan  et  al.,  2015;  Neidell  et  al.,  2014;  Gallino  et  al.,  2012;  Hsiang,  2010).    

 q  Establishing  a  causal  link  between  temperature  shocks  and  firm  behavior    

in  climate  vulnerable  countries    will  significantly  improve  the  design  of  climate  change  policy.  

Evidence  of  Warming  in  CI  

 

DECADE  

Average  Temperature  (°C)    

 

Normal  Period:                      1961-­‐1990  (25,7°C)  

   1961-­‐1970     25,5     -­‐0,2    

1971-­‐1980     25,6     -­‐0,1    

1981-­‐1990     25,9     0,2    

1991-­‐2000     26,1     0,4    

2001-­‐2010     26,5     0,8    

Ø   Warming  of  about  0.5°C  since  the    decade  of  1980    Ø Difference  of  +1.2°C  compare  to  the  normal  of  (1961-­‐1990)  

Source:  SODEXAM/DirecIon  de  la  Météorologie  NaIonale  de  Cote  d'Ivoire  

 

q   Average  temperature  in  CI:  1961-­‐2010  

Evidence  of  Warming  in  CI  

21.0  

22.0  

23.0  

24.0  

25.0  

26.0  

27.0  

28.0  

29.0  

Jan   Fév   Mar   Avr   Mai   Jun   Jul   Aoû   Sep   Oct   Nov   Déc  

Tempé

rature  (°C)  

Mois  

Normale  1961-­‐1990  

Normale  1971-­‐2000  

Normale  1981-­‐2010  

Source:  SODEXAM/DirecIon  de  la  Météorologie  NaIonale  de  Cote  d'Ivoire  

Ø  Maxi  :  Feb-­‐April  Ø  Mini:  Jul-­‐Sept    

q   Trend  of  Average  Annual  Temperature  in  Abidjan  

Evidence  of  Warming  in  CI  

280

300

320

340

360

num

days

_25

1980 1990 2000 2010year

kernel = epanechnikov, degree = 0, bandwidth = 2.55

Local polynomial smooth

q   Temperature  in  CI:1980-­‐2010  

DistribuIon  of  Firm  Output  

0.0

5.1

.15

.2.2

5

0 5 10 15x

kdensity logproduction_exit kdensity logproduction_survkdensity logproduction

Literature  Review  

q Melitz  (2003):  Exposure  to  trade  induces  the  most  producIve  firm  to  enter  the  export  market  and  forces  the  least  producIve  firm  to  exit.  

 q Macro  studies  of  the  impact  of  temperature  on  economic  acIviIes:  

Burke  et  al.  (2015),  Dell  et  al.  (2012),  Hsiang  (2010).      q Micro-­‐level  studies  of  the  impact  of  temperature  on  economic  

acIviIes:  Park  (2016);  Sunarshan  et  al.  (2015);  Neidell  et  al.  (2014);  and  Gallino  et  al.(  2012)  

 q Evidence  of  the  effects  of  temperature  on  labor  producIvity  and  

supply:  Park  (2016);  Sunarshan  et  al.  (2015);  Neidell  et  al.  (2014);  Gallino  et  al.,  (2012);  Hsiang  (2010)    

   

Model:  PotenIal  Channels  of  transmission    

 q Labor  producIvity/Supply  

o  High  temperatures  cause  discomfort,  faIgue,  cogniIve  impairment  o  Disease,  leisure  

q ProducIon  cost/operaIng  costs    o  Power  disrupIon,  installing  AC,  buying  generators  o  Input  prices  

   

Model  

             q  Setup    

 

 

§ Assuming  that  firms  face  exogenous  constant  elasticity  of  

substitution  (CES)  demand  schedule,  optimal  quantities  produced  

by  a  representative  firm  are  determined  by:  

 

(2)      qi = (piP+−σQ    

 

where  

pi −  respectively  firm’s  price,  P −    aggregate  price  index,  Q −  aggregate  demand,  and  σ −    elasticity  of  substitution  between  in  any  two  goods  

   

Model  

q  Setup      

q  Firm’s  Entry      

§ The  economy  is  populated  with  a  continuum  of  monopolistically  competitive  firms,   each   producing   a   differentiated   product   under   increasing   return   to  scale  technology.    

 § In  sector  j,  each  variety  is  produced  using    labor  Lj  ,  with  unit  cost  wj ,  as  the  

sole  factor  of  production.    

§ Within   the   industry,   firms   are   heterogeneous   in   their   productivity  φ,   and  they  face  fixed  sector  entry  cost  of    fj  units  of  labor.  

 § Productivity  φ,  unknown  to  firms  before  starting  production,  is  drawn  from  a  

known  cumulative  Pareto  distributive  function  G(φ).    

 

Model  

q  Setup      

q  Technology    § After  observing  their  productivity,  firms  decide  whether  to  stay  or  exit  the  

market.    Upon  staying  in  the  market,  firms  decide  how  much  to  produce  and  

assess  whether  to  invest  in  climate  adaptation  technologies.    

§ Production  of  each  variety  involves  fixed  cost  (fj)  and  a  marginal  cost  #1 φ% &,  

both  in  terms  of  units  of  labor.    

 

§ Assumption  1:  temperature  shocks  affect  the  production  of  firms  in  all  

sectors  within  the  industry.  The  vulnerability  of  firms  to  temperature  shocks  

varies  across  sectors.  

   

Model  

q  Setup    q  Climate  AdaptaIon  Technology      

 

 

§ By  investing  in  climate  adaptation  technology  at  fixed  cost  θj  units  of  labor,  a  

firm  reduces  its  marginal  labor  input  cost  by  𝜏,  with  𝜏 < 1.    The  total  cost  of  

producing  qjunits  of  a  variety  is:      

(2)  lj =  fj +qjφ− e0𝜏lj − θj1  

 

where  𝑙  𝑗  is   labor   used   for  𝑞  units   of   a   variety   in   sector    𝑗  and  𝑒  is   a   binary  

indicator  which   is   equal   to  one   if   firm  makes   climate   adaptation   technology  

investment  and  zero  otherwise.  

 

Assumption  2:  the  fixed  cost  of  investing  in  climate  adaptation  technology  (𝜃𝑗 )  

increases  in  temperatures  and  varies  by  sectors.    

 

 

Timeline  of  Events  

t   t+1  

Firm  observes  its    producIvity    

Choose  q      and  e    

ProducIon    starts  

Revenues    are  realized  

Model  

q  Firm  Behavior      § Under  CES  preferences  and  in  the  absence  of  climate  adaptation  technology  

investment,  the  static  problem  for  a  monopolistic  competitive  firm,  with  

productivity  𝜑,  is  to  maximize  the  following  profit  function:    

 

(3)      π(φ) = maxl,q

pq −  wl − wf  

s. t.  

(4)      l =qφ  

(5)      qi = 7piP9−σQ  

 

 

Model  

q  Firm  Behavior  q  Equilibrium     § From  the  first  order  condition  for  profit  maximization,  the  equilibrium  price  

for  each  variety  is  a  constant  mark-­‐up  over  marginal  cost:  

 

(6)    𝜎

𝜎 − 1𝑤𝜑  

which  results  in  an  equilibrium  firm  revenue  of:  

 

(7)    𝑟(𝜑) = 𝐴 .𝜎−1𝜎/𝜎−1

𝑤1−𝜎𝜑𝜎−1,    with  𝐴 = 𝑄𝑃𝜎  

and  a  corresponding  equilibrium  profit  of:  

 

(8)      𝜋(𝜑) =𝑟(𝜑)𝜎

− 𝑤𝑓 = 𝐵𝜑𝜎−1 − 𝑤𝑓, 𝑤𝑖𝑡ℎ  𝐵 = 𝐴(𝜎 − 1)𝜎−1

𝜎𝜎𝑤1−𝜎  

 

 

Model  

q Equilibrium        

     

§ A  monopolistic  competitive  firm  which  invest  in  climate  adaptation  technology  maximizes:    

 

(9)      π(φ) = maxl,q

pq −  wl(1 − 𝜏) − w(f + θ)  

s. t.  equation  (2)  and  (5)  hold.    

From  the  first  order  condition,  we  get  the  following  equilibrium  price  for  each  variety:  

 

(10)    𝜎

𝜎 − 1(1 − 𝜏)𝑤

𝜑  

which  implies  an  equilibrium  firm  revenue  of:  

 

(11)      𝑟(𝜑) = 𝐴 <𝜎−1𝜎=𝜎−1

[(1 − 𝜏)𝑤]1−𝜎𝜑𝜎−1,    with  𝐴 = 𝑄𝑃𝜎  

and  an  equilibrium  profit  of:  

 

(12)  𝜋(𝜑) =𝑟(𝜑)𝜎

− w(f + θ) = 𝐵ʹ′𝜑𝜎−1 − w(f + θ), 𝐵ʹ′ = 𝐴(𝜎 − 1)𝜎−1

𝜎𝜎[(1 − 𝜏)𝑤]1−𝜎  

   

Model  

q  Firm’s  Behavior    q  Firm’s  Decision      § Firms  make  the  joint  decision  of  whether  to  stay  in  the  market  and  whether  

to  invest  in  climate  adaptation  technology  comparing  the  profit  in  equation  

(8)  and  (12)  to  zero  and  to  each  other.    

 

§ A  representative  firm  after  making  the  decision  to  produce  will  only  invest  in  

climate  adaptation  technology  if  the  profit  in  equation  (12)  is  greater  than  

the  profit  in  equation  (8).  In  other  words,  a  firm  invests  in  climate  adaptation  

technology  only  if  𝜏l ≥ 𝜃.  

 

§ Proposition  1:  for  a  given  𝜑  𝑎𝑛𝑑  𝜏,  there  exit  a  level  of  temperature  beyond  

which  it  is  always  optimal  to  invest  in  climate  adaptation  technology.      

   

Model  

q  PredicIons    

   

 

Proposition  2:  The  solution  to  the  firm  profit  maximization  problem  solve  the  

productivity  level  threshold,  𝜑∗,  such  that  a  representative  firm  stay  in  the  

market  only  if  its  productivity  level  is  greater  than  𝜑∗.  

 

(13)    𝜑∗ = )*1

𝜎 − 1-𝑤(1 − 𝜏)𝑃𝑄1 𝜎2

3

𝜎𝜎−1

*𝜃𝜏-

1𝜎−1

 

 

The  model  outlined  in  this  section  generates  two  testable  implications  of  

temperature  shocks,  which  will  be  evaluated  in  the  empirical  section:  

 

Implication  1:  Under  Assumption  1-­‐2,  Proposition  2  implies  that  an  increase  

in  temperatures  increases  the  cutoff  productivity  level  and,  consequently,  

increases  firm's  exit  rate.    

 

 

Data  

           We  use  two  sources  of  data  to  analyze  the  impact  of  temperature  shocks  and  volaIlity  on  firm’s  dynamics:  

   1.  Firm-­‐level  data  from  the  Registre  des  Entreprises  de  Cote  d’Ivoire,  

covering  the  universe  of  registered  firms  in  Cote  d’Ivoire  from  1996-­‐2012.  The  data  is  collected  by  the  NaIonal  InsItute  of  StaIsIcs  (INS).      

§  The  data  contains  informaIon    on  output,  inputs,  sales  (domesIc,  and  exported),  employment,  ownership  status,  and  operaIng  costs  of  all  formal  agricultural,  manufacturing,  service,  and  trade  establishments  in  the  country  

 2.  Temperature  and  precipitaIon  data  from  NASA  1980-­‐2012  daily  Ime  

series  -­‐  0.5X0.5  degree  resoluIon.    

   

 

Preliminary  Summary  StaIsIcs  

Table&1a:&Panel&length

YEAR NUMBER&OF&FIRMS1996 2701997 3071998 2361999 3622000 3402001 3832002 3752003 3252004 3382005 3762006 3672007 3272008 4002009 4272010 4972011 3862012 737TOTAL 6453

Note:&firms&from&nonAmanufacturing&sectors&are&removed

Preliminary  Summary  StaIsIcs  

Table&1b:&Subsector&Length

YEAR NUMBER&OF&FIRMSTextiles(and(Leather(Products 531Chemical(Products 647Edition(and(Printings 789Food(Products 1486Construction(Equipment( 194Communication(Equipment( 201Metallurgy(and(Furniture 864Plastic(and(Rubber(Products 744Vehicle(and(boat 101Wood(and(Carton( 896TOTAL 6453

Summary  StaIsIcs  

Table&2:&Summary&Statistics

VARIABLES AVERAAGE SD

FIRM&VARIABLESOutput%value 6,560 41,432Number%of%Employee 154.7 627Wage%bill 497.5 1,433Investment 374.9 1,721Land 4.054 258.1Water%consumption 83.12 719.3Energy%Consumption 216 624.9Intermediate%input 4,929 38,749Tax 207.4 2,551R&D 37.26 353.5Subsidy 5.716 130.7Other%revenue 2.434 55.07Firm%age 15.3 12.94

CLIMATE&VARIABLESAverage%temperature%in%°C 27.07 0.326Number%of%days%above%27°C 166.2 40.41Number%of%days%above%30°C 1.951 6.128Average%rain%in%mm 1,757 215.6

Observation%# 6,453

Note:&currency&is&in&thousand&of&CFA&D&$1&~550&CFA

Empirical  Framework  

q   Effects  of  temperature  on  producIvity  

 

 

Our  empirical  strategy  for  studying  the  impact  of  temperature  shocks  on  firm's  

performance  is  to  regress  firm's  total  factor  production  (TPF)  measures,  estimated  

using  the  GNR  approach,  on  the  temperature  shocks  and  other  control  variables.  

Our  key  regression  is  thus  specified  as:  

 (14)    log(𝑇𝑃𝐹𝑖𝑡 ) = 𝛾𝑖 + 𝛾𝑡 + 𝛾1𝑙𝑜𝑔𝑡𝑒𝑚𝑝𝑖𝑡 + 𝛾2𝑙𝑜𝑔𝑟𝑎𝑖𝑛𝑖𝑡 + 𝑋𝑖𝑡ʹ′ 𝛽 + 𝜀𝑖𝑡      where  TPFit  is  firm’s  i  total  factor  productivity  in  year  t,γi  and  γt  control  

respectively  for  firm  and  year  fixed  effects,  tempit  measures  temperature  shocks  

and  volatility  at  firm’s  i  location  in  year  t,  𝑟𝑎𝑖𝑛𝑖𝑡  is  total  rain  firm’s  i  location  in  

year  t,    Xit  production  factors  and  firm’s  characteristics,  and  εit  is  the  error  term.  

Empirical  Framework  

q  Effect  of  Temperature  on  Firm’s  Exit  

Unlike  previous  works,  we  extend  the  analysis  by  also  looking  at  the  implication  of  

the  effects  of  temperature  shocks  on  firm’s  exit  by  estimating  the  following  

equation:  

   (15)    Probit(Exitit ) = Φ(α0 + α1tempit + α1ageit + 𝑋𝑖𝑡ʹ′ 𝛿 + 𝜖𝑖𝑡 )  

 

where  Exitit  is  the  dummy  variable  for  whether  firm  is  active  in  year  t  or  not,  

tempit  and  ageit  denotes  respectively  the  temperature  variable  and  the  age  of  the  

firm,  and  Xitare  controlled  covariates.  

 

Preliminary  Results  

Table&3:&Effects&of&temperature&on&firm&output

VARIABLES

(1),,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,Log,production,

(2),,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,Log,production,

(3),,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,Log,production,

(4),,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,Log,production,

Log,number,of,days,above,30°C E0.0261* E0.0342**

(0.0141) (0.0141)Log,yearly,total,rain E0.352*** E0.0742 E0.0805 E0.333***

(0.0881) (0.0684) (0.0694) (0.0884)Log,production,cost 0.910*** 0.885*** 0.885*** 0.895***

(0.0136) (0.00845) (0.00844) (0.0183)Log,number,of,employee 0.0277* 0.0310*** 0.0310*** 0.0239

(0.0164) (0.0103) (0.0103) (0.0195)Log,capital 0.0171* 0.0259*** 0.0259*** 0.0190**

(0.00913) (0.00589) (0.00589) (0.00822)Log,energy,and,water 0.0738*** 0.0730*** 0.0728*** 0.0881***

(0.0144) (0.00823) (0.00820) (0.0150)

Log,number,of,days,below,27°C 0.0303

(0.0382)

Log,yearly,average,temperature E0.626

(0.770)Constant 2.934*** 0.795 3.060 2.857***

(0.663) (0.533) (2.680) (0.651)

Firm&Random&Fixed&Effects No No No Yes

Observations 1,337 4,708 4,708 1,337REsquared 0.939 0.911 0.911

Note:,Standard,errors,in,parentheses.,***,p<0.01,,**,p<0.05,,*,p<0.1.,,The,average,temperature,is,about,27°C.,The,number,of,days,below,the,average,temperature,has,a,positive,but,insignificant,effects,on,productivity.,

Preliminary  Results  

Table&4:&Mechanisms&of&transmission&of&temperature&effects

VARIABLES(1),,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,

Log,production,(2),,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,

Log,production,

Log,number,of,days,above,30°C 0.0653 G0.0157(0.0483) (0.0178)

Log,yearly,total,rain G0.169 G0.263***(0.111) (0.0992)

Log,number,of,employee G0.0758** G0.000632(0.0373) (0.0229)

Log,capital 0.0466*** 0.0258**(0.0175) (0.0117)

Log,production,cost 0.892***(0.0199)

Log,energy,&,water,consumption 0.0776*** 0.0922***(0.0280) (0.0177)

Food,manufacturing 0.0122(0.0675)

Food,manufacturing,x,temperature 0.00127

(0.0288)Vehicle,manufacturing 0.160

(0.194)

Vehicle,manufacturing,x,temperaure G0.265***

(0.0734)Log,firm,age 0.0311 0.00727

(0.0290) (0.0205)Log,number,of,employee,x,temperature 0.0213

(0.0167)Log,wage 0.349***

(0.0394)

Log,wage,x,temperature G0.0359*

(0.0189)

Log,energy,x,temperature 0.0204*

(0.0121)

Log,age,x,temperature G0.0256**

(0.0128)Log,input 0.601***

(0.0195)

Observations 1,029 1,029Note:,Standard,errors,in,parentheses.,***,p<0.01,,**,p<0.05,,*,p<0.1.,Except,food,manufacturing*temperature,,all,other,nonGsignificant,interactions,are,removed.,Log,of,number,of,days,above,30°C,is,used,as,the,temperature,interaction,variable.,

Preliminary  Results  

Table&5:&Effects&of&temperature&on&firm's&exit

Variables Prob.&Of&Exit Prob.&Of&Exit

Number'of'days'above'30°C 0.0377*** 0.0348**(0.0107) (0.0142)

Number'of'days'above'30°C'squared =0.000814** =0.00142***(0.000342) (0.000393)

Yearly'total'rain's =0.00205(0.00209)

Yearly'total'rain'squared 3.56e=07(5.81e=07)

Firm'Age =0.00291**(0.00144)

Number'of'employee =0.000131***(4.58e=05)

capital 1.18e=05(1.03e=05)

Energy'and'water 8.50e=05***(2.28e=05)

Observations 6,453 4,802Notes:&Standard&errors&in&parentheses.&***&p<0.01,&**&p<0.05,&*&p<0.1

Conclusion  

q  We  have  argued  that  higher  temperatures  affect  firm’s  producIvity    q  Our  preliminary  results  is  aligned  with  the  predicIon  of  the  model  we  

developed  by  introducing  climate  adaptaIon  technology  cost  into  Melitz  (2003)  model.    

 q  We  find  that  a  one-­‐percent  increase  in  the  number  of  days  with  average  

temperatures  above  30°C  is  associated  with  a  0.034%  decreases  in  firm’s  output.  An  extra  day  with  average  temperature  above  30°C  is  associated  with  3.48%  increase  in  firm’s  likelihood  to  exit.  

 q  We  also  find  that  an  extra  day  of  temperature  above  30°C  decreases  

intrinsic  labor  producIvity  by  3.59%.