sessionv dickens handout

What effects is EMU having on the euro area and its member countries? 16 – 17 June 2005, Frankfurt am Main

Session V: Inflation persistence and inflation differentials Discussant:

William Dickens Brookings Institution, Washington D.C.

1

Comments on “Price Setting Comments on “Price Setting and Inflation Persistence: did and Inflation Persistence: did

EMU Matter?”EMU Matter?”

European Central BankEuropean Central Bank17 June, 200517 June, 2005

Frankfurter Frankfurter HofHof HotelHotel

What I Want to DoWhat I Want to Do

Heap praise on the authors and the Heap praise on the authors and the organizers of the IPN.organizers of the IPN.Go over results of Go over results of AngeloniAngeloni et. al. et. al. Express some concerns about the results Express some concerns about the results and their interpretation.and their interpretation.Present some preliminary results from the Present some preliminary results from the International Wage Flexibility Project (a International Wage Flexibility Project (a sister project of IPN which has also been sister project of IPN which has also been generously supported by the ECB).generously supported by the ECB).

2

This is Hard WorkThis is Hard Work

Perhaps no one in this room knows better than I Perhaps no one in this room knows better than I do how much work goes into a project like the do how much work goes into a project like the IPN. IPN. Until you have tried to coordinate 13 teams of Until you have tried to coordinate 13 teams of researchers in 12 countries dealing with more researchers in 12 countries dealing with more than a dozen unique sets of data and tried to than a dozen unique sets of data and tried to administer a common protocol and make sense administer a common protocol and make sense of the results…of the results…So my hat is off to the authors. Been there, done So my hat is off to the authors. Been there, done that, its not easy! that, its not easy!

AngeloniAngeloni et. al. Presentet. al. PresentTwo Types of EvidenceTwo Types of Evidence

Micro data on frequency and magnitude of Micro data on frequency and magnitude of price changes from surveys of consumer price changes from surveys of consumer and producer pricesand producer pricesRegression estimates of inflation Regression estimates of inflation persistence (sum of coefficients in a persistence (sum of coefficients in a regression of inflation on many lags) using regression of inflation on many lags) using product level inflation data.product level inflation data.

3

Did EMU Matter?Did EMU Matter?Major FindingsMajor Findings

PRICE CHANGES: PRICE CHANGES: No. Other than an abrupt No. Other than an abrupt and limited increase in the frequency of price and limited increase in the frequency of price changes around the time of cash conversion changes around the time of cash conversion there is no evidence of any change in the there is no evidence of any change in the frequency or magnitude of price changes.frequency or magnitude of price changes.INFLATION PERSISTENCE:INFLATION PERSISTENCE: Some evidence Some evidence that inflation persistence declined in mid 90s, but that inflation persistence declined in mid 90s, but same thing happened in US and likely reflects same thing happened in US and likely reflects change in nature of inflationary expectations change in nature of inflationary expectations (which is likely due in some sense to EMU).(which is likely due in some sense to EMU).

Why Do We Care About Frequency Why Do We Care About Frequency of Price Changes?of Price Changes?

What I would most like to see is reference to a What I would most like to see is reference to a model of firm behavior that would allow me to model of firm behavior that would allow me to judge whether the estimated frequency is judge whether the estimated frequency is “high” or “low” relative to some policy relevant “high” or “low” relative to some policy relevant benchmark. Ideally what would be reported benchmark. Ideally what would be reported and tested for change would be parameters and tested for change would be parameters such as implied cost of adjusting and/or not such as implied cost of adjusting and/or not adjusting prices.adjusting prices.But, at the very least it would be good if But, at the very least it would be good if measures of frequency of change were measures of frequency of change were adjusted for inflation rates.adjusted for inflation rates.

4

Adjust Change Frequencies Adjust Change Frequencies for Inflation?for Inflation?

We might expect the frequency of price changes We might expect the frequency of price changes to decline considerably with declining to decline considerably with declining inflation.inflation.What I think we care about (and without an What I think we care about (and without an explicit model I’m not sure) is whether EMU (and explicit model I’m not sure) is whether EMU (and the single market more broadly) affected the the single market more broadly) affected the costs of changing, or not changing, prices. costs of changing, or not changing, prices. The fact that there has been no change in the The fact that there has been no change in the frequency of price reductions in countries that frequency of price reductions in countries that have had significant drops in their rates of have had significant drops in their rates of inflation may indicated an increase in price inflation may indicated an increase in price flexibility.flexibility.

Why Do We Care About Why Do We Care About Persistence?Persistence?

Persistence measures are tied to new Persistence measures are tied to new Keynesian Phillips curve and as such are Keynesian Phillips curve and as such are directly connected with policy concerns (how directly connected with policy concerns (how quickly will shocks abate and how aggressive quickly will shocks abate and how aggressive must policy be to stabilize inflation and output).must policy be to stabilize inflation and output).But link is tenuous due to nature of estimates But link is tenuous due to nature of estimates (“reduced form” auto(“reduced form” auto--regressions rather than regressions rather than estimates of models).estimates of models).Thus we don’t know how persistent persistence Thus we don’t know how persistent persistence is (could be structural or ephemeral ). is (could be structural or ephemeral ).

5

We May Not Believe Model We May Not Believe Model Estimates…Estimates…

…… but at least they give us some breakdown but at least they give us some breakdown of the source of persistence. (Is it due to past of the source of persistence. (Is it due to past inflation affecting current inflation, the way inflation affecting current inflation, the way expectations are formed, or autoexpectations are formed, or auto--correlation correlation in real or nominal innovations?)in real or nominal innovations?)I would be remiss as a discussant if I didn’t I would be remiss as a discussant if I didn’t take this opportunity to mention my favorite take this opportunity to mention my favorite model of this phenomena model of this phenomena –– AkerlofAkerlof, Dickens , Dickens and Perry (2000).and Perry (2000).

ADP 2000 and PersistenceADP 2000 and Persistence

Our model predicts the decline in “persistence” Our model predicts the decline in “persistence” as inflation falls as the result of people as inflation falls as the result of people increasingly ignoring expectations in price increasingly ignoring expectations in price setting with lower inflation.setting with lower inflation.We find that once we take account of this the We find that once we take account of this the parameters of our Phillips Curve for US data are parameters of our Phillips Curve for US data are quite stable.quite stable.Our model suggests that an inflation target of Our model suggests that an inflation target of about 3% minimizes unemployment for the US about 3% minimizes unemployment for the US (model doesn’t fit well for European data).(model doesn’t fit well for European data).

6

The International Wage The International Wage Flexibility ProjectFlexibility Project

What is the IWFP?What is the IWFP?

13 Country study of wage inflation 13 Country study of wage inflation supported by the ECB and directed by supported by the ECB and directed by Erica Erica GroshenGroshen (NY Fed.) and me(NY Fed.) and meusing micro data on individual and using micro data on individual and occupational wages analyzed by teams occupational wages analyzed by teams in each country familiar with the data in each country familiar with the data broader than analyzing wage rigidity broader than analyzing wage rigidity (sand and grease), but that is the part (sand and grease), but that is the part that I’m going to talk about todaythat I’m going to talk about today

7

Country TeamsCountry Teams

AustriaAustriaBelgiumBelgiumDenmarkDenmarkFinlandFinlandFranceFranceGermanyGermanyItalyItaly

NorwayNorwayPortugalPortugalSwedenSwedenSwitzerlandSwitzerlandUnited KingdomUnited KingdomUnited StatesUnited States

DisclaimersDisclaimers

PreliminaryPreliminary results! Mainly based on results! Mainly based on European Community Household Panel European Community Household Panel data which is quite noisy.data which is quite noisy.My views only (not the views of the ECB, My views only (not the views of the ECB, Brookings, or any other members of the Brookings, or any other members of the IWFP project team).IWFP project team).

8

QuestionsQuestionsCould DNWR be an explanation for Europe’s Could DNWR be an explanation for Europe’s persistent unemployment problems or…persistent unemployment problems or…While ADP model fits very well for US and While ADP model fits very well for US and Canadian data it fits very poorly and gives Canadian data it fits very poorly and gives nonsensical results when applied to European nonsensical results when applied to European data. Could presence of real rigidity caused by data. Could presence of real rigidity caused by centralized wage setting institutions be the centralized wage setting institutions be the reason for this?reason for this?Some studies done in Europe (notably UK) show Some studies done in Europe (notably UK) show much less Downward Nominal Wage Rigidity much less Downward Nominal Wage Rigidity than US and Canadian data, so is DNWR a than US and Canadian data, so is DNWR a problem in Europe? Might problem be downward problem in Europe? Might problem be downward real rigidity instead?real rigidity instead?

How Could We Answer Questions?How Could We Answer Questions?

Initially we were unsure about how to get Initially we were unsure about how to get at presence of different types of rigidity.at presence of different types of rigidity.At first meeting some very interesting At first meeting some very interesting results emerged examining wage change results emerged examining wage change histograms for job histograms for job stayersstayers (all results are (all results are for job for job stayersstayers).).

9

Fra

ctio

n

Does Low Inflation Eliminate Downward Wage RigiThe Distribution of Nominal Wage Changes, Switzerland 1986 - 199

1986, Inflation: 0%

0

.05

.1

.15

.2

1987, Inflation: 1.9% 1988, Inflation: 1.9% 1989, Inflation: 5%

1990, Inflation: 5.3%

0

.05

.1

.15

.2

1991, Inflation: 5.2% 1992, Inflation: 3.4% 1993, Inflation: 2.5%

1994, Inflation: .4%

0

.05

.1

.15

.2

1995, Inflation: 1.9% 1996, Inflation: .8%

-.25 -.1 0 .1 .25

1997, Inflation: .4%

-.25 -.1 0 .11998, Inflation: -0.2%

-.25 -.1 0 .1 .250

.05

.1

.15

.2

1999, Inflation: 1.7%

-.25 -.1 0 .1 .25

Growth Rate 1978-1979

0

50,000

100,000

150,000

200,000

250,000

300,000

350,000

400,000

[-20,-1

9[

[-17,-1

6[

[-14,-1

3[

[-11,-1

0[[-8

,-7[

[-5,-4

[[-2

,-1[

[1,2[[4,5[

[7,8[

[10,11[[13,14[

[16,17[[19,20[

Growth Rate

N an

d %

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

Belgium

10

Growth Rate 1981-1982

0

50,000

100,000

150,000

200,000

250,000

300,000[-2

0,-1

9[[-1

8,-1

7[[-1

6,-1

5[[-1

4,-1

3[[-1

2,-1

1[[-1

0,-9

[[-8

,-7[

[-6,-5

[[-4

,-3[

[-2,-1

[[0

,1[

[2,3

[[4

,5[

[6,7

[[8

,9[

[10,

11[

[12,

13[

[14,

15[

[16,

17[

[18,

19[

Growth Rate

N a

nd %

0

0.02

0.04

0.06

0.08

0.1

0.12

Fra

ction

1984

-15 -10 -5 0 5 10 15 20 25 30 35

0

.05

.1

.15

UK

11

Can we use wage change Can we use wage change histograms to diagnose nature histograms to diagnose nature

and extent of wage rigidity?and extent of wage rigidity?

Wage Change Distribution(For workers with same wage facing same minimum wage)

-0.22-0.19-0.16-0.13-0.10-0.07-0.04-0.010.020.050.080.110.140.170.200.230.26

Change in Log Wage

Freq

uenc

y

No Nominal Wage Change

No Real Wage Change

Wage Reduced to Minimum

Observatuions Swept Down to Zero by

Menu Costs

Observations Swept Up to Rate of Price Inflation By Downward Real

Wage Rigidity

Observations Swept Up to Zero By Downward Nominal Wage Rigidity

and Menu Costs

12

Big Problem is Measurement Big Problem is Measurement ErrorError

If people make mistakes reporting their wages then If people make mistakes reporting their wages then we see wage changes where there are none.we see wage changes where there are none.If we compute wage=income/hours we see “wage” If we compute wage=income/hours we see “wage” changes due to overtime, bonuses, or mistakes in changes due to overtime, bonuses, or mistakes in reporting hours.reporting hours.If we use social security data we have similar If we use social security data we have similar problems since most countries don’t have data that problems since most countries don’t have data that allow us to accurately identify base wage.allow us to accurately identify base wage.All evidence suggests that for most data sets All evidence suggests that for most data sets frequency and extent of errors make this a very frequency and extent of errors make this a very serious problem (evidence suggests that in many serious problem (evidence suggests that in many data sets most reported wage cuts are actually data sets most reported wage cuts are actually errors of these sorts).errors of these sorts).

New Approach:New Approach:Use information in correlation Use information in correlation

of changes between yearsof changes between yearsAbowdAbowd and Card (1989) suggest that wages have and Card (1989) suggest that wages have two components:two components:

permanent changespermanent changestransient (one period) changes (which result in transient (one period) changes (which result in negative serial correlation of wage changes)negative serial correlation of wage changes)

New method identifies transient changes as errors New method identifies transient changes as errors and uses autoand uses auto--covariance and frequency of sign covariance and frequency of sign switching in changes to identify error rate and error switching in changes to identify error rate and error variance.variance.This information is used to identify semiThis information is used to identify semi--nonnon--parametric mixed method of moments estimate of parametric mixed method of moments estimate of true wage distribution. true wage distribution.

13

Rigidity MeasuresRigidity MeasuresEstimated true wage change distributions are Estimated true wage change distributions are then used as data for GMM estimate of a simple then used as data for GMM estimate of a simple model of three types of rigiditymodel of three types of rigidity

Downward nominal wage rigidityDownward nominal wage rigiditySymmetric nominal wage rigidity (menu costs)Symmetric nominal wage rigidity (menu costs)Downward real wage rigidityDownward real wage rigidity

Measure is fraction of notional wage cuts (or Measure is fraction of notional wage cuts (or small changes) that don’t happen due to rigidity small changes) that don’t happen due to rigidity (so could be independent of rate of inflation).(so could be independent of rate of inflation).No measures of insensitivity to fundamentals No measures of insensitivity to fundamentals (another concept of rigidity).(another concept of rigidity).

Validating New Validating New Error Correction MethodError Correction Method

Results applying new method to US largely fit with Results applying new method to US largely fit with those of other studies (a very high degree or those of other studies (a very high degree or downward nominal rigidity).downward nominal rigidity).Gottschalk method identifies true wages and Gottschalk method identifies true wages and measurement error in quarterly data. True wage measurement error in quarterly data. True wage changes have autochanges have auto--correlation of zero (our key correlation of zero (our key assumption).assumption).Finnish data have very few errors and procedure Finnish data have very few errors and procedure makes very few correctionsmakes very few correctionsPortuguese have good and bad data. Procedure Portuguese have good and bad data. Procedure doesn’t correct good data and makes distribution of doesn’t correct good data and makes distribution of bad data resemble the good data.bad data resemble the good data.

14

0.05

.1.1

5.2

-4 -3 -2 - 1 0 1 2 3 4 5 6 7 8 9 10 1 1 1 2 1 3 14 15 16 17 18 19

1971

observed distribution estimated true wage distributionconstrained distribution

0.0

5.1

.15

-4 -3 -2 - 1 0 1 2 3 4 5 6 7 8 9 10 1 1 1 2 1 3 14 15 16 17 18 19

1976


0.0

5.1

.15

-4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 1 3 14 15 16 17 18 19

1981


0.1

.2.3

-4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 1 3 14 15 16 17 18 19

1992


US PSID Error Correction Model0

.05

.1.1

5.2

-4 -3 -2 - 1 0 1 2 3 4 5 6 7 8 9 10 1 1 1 2 1 3 14 15 16 17 18 19

1991


0.1

.2.3

.4

-4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 1 3 14 15 16 17 18 19

1998


0.2

.4.6

.8

-4 -3 -2 - 1 0 1 2 3 4 5 6 7 8 9 10 1 1 1 2 1 3 14 15 16 17 18 19

1993


0.2

.4.6

-4 -3 -2 - 1 0 1 2 3 4 5 6 7 8 9 10 11 12 1 3 14 15 16 17 18 19

2001


Finland Service Sector

15

Portugese Data

Base wage+other labor earnings -1994

-0.02

0

0.02

0.04

0.06

0.08

0.1

0.12

-25 -23 -21 -19 -17 -15 -13 -11 -9 -7 -5 -3 -1 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49

Base wage - 1994

-0.02

0

0.02

0.04

0.06

0.08

0.1

0.12

-25 -23 -21 -19 -17 -15 -13 -11 -9 -7 -5 -3 -1 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49

Base wage+other labor earnings-1992

-0.01

0.01

0.03

0.05

0.07

0.09

0.11

0.13

0.15

-25 -22 -19 -16 -13 -10 -7 -4 -1 2 5 8 11 14 17 20 23 26 29 32 35 38 41 44 47 50

Base wage-1992

-0.01

0.01

0.03

0.05

0.07

0.09

0.11

0.13

0.15

-25 -22 -19 -16 -13 -10 -7 -4 -1 2 5 8 11 14 17 20 23 26 29 32 35 38 41 44 47 50

From ECHP and PSIDReal and Nominal Rigidity in 12 Countries

-0.1

0.1

0.3

0.5

0.7

0.9

1.1

Denmark

Greece US

Italy

Netherl

andsAus

tria UK

German

yIre

land

Finlan

d

France

Belgium

Frac

tion

Affe

cted

Real Nominal

16

Real and Nominal Rigidity in 10 EMU Countries

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

1995 1996 1997 1998 1999 2000 2001

Rig

idity

Real

Nominal

Dispersion of Wage Rigidity in the EMU

0.20.220.240.260.280.3

0.320.340.360.38

1995 1996 1997 1998 1999 2000 2001

Rigi

dity

SD Real Rigidity SD Nominal Rigidity

17

ConclusionsConclusions

Too early to conclude much, but…Too early to conclude much, but…Much of Europe is different from the US showing Much of Europe is different from the US showing few signs of DNWR and substantial evidence of few signs of DNWR and substantial evidence of DRWR.DRWR.Real and nominal rigidity seem to be negatively Real and nominal rigidity seem to be negatively related and there are both high and low real related and there are both high and low real rigidity countries in EMUrigidity countries in EMUPreliminary evidence suggests that wage setting Preliminary evidence suggests that wage setting institutions (centralized collective bargaining) are institutions (centralized collective bargaining) are important determinants of regime. important determinants of regime.

sessionv dickens handout

Documents