productivity differentials in the u.s. and eu distributive ... · germany 1.4 -1.0 2.7 1.6 2.2 1.3...

19
1 University of Groningen Productivity Differentials in the U.S. and EU Distributive Trade Sector: Statistical Myth Or Reality? Marcel Timmer Robert Inklaar Bart van Ark University of Groningen and The Conference Board OECD Workshop on Services 15 & 16 November 2004

Upload: others

Post on 24-Jun-2020

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Productivity Differentials in the U.S. and EU Distributive ... · Germany 1.4 -1.0 2.7 1.6 2.2 1.3 Netherlands 2.4 1.3 2.1 3.5 2.5 1.1 ... Changing business models in distributive

1

University of Groningen

Productivity Differentials in the U.S. and EU Distributive Trade Sector:

Statistical Myth Or Reality?

Marcel TimmerRobert InklaarBart van Ark

University of Groningenand The Conference Board

OECD Workshop on Services15 & 16 November 2004

Page 2: Productivity Differentials in the U.S. and EU Distributive ... · Germany 1.4 -1.0 2.7 1.6 2.2 1.3 Netherlands 2.4 1.3 2.1 3.5 2.5 1.1 ... Changing business models in distributive

2

BackgroundGrowth of value added per hour worked in trade, U.S. and EU

1980-1995 1995-2002 1980-1995 1995-2002 1980-1995 1995-2002U.S. 2.0 4.3 4.0 8.1 2.5 7.1EU-15 1.9 1.2 2.7 1.6 2.0 1.5

France 1.9 0.5 4.7 1.2 3.4 1.5Germany 1.4 -1.0 2.7 1.6 2.2 1.3Netherlands 2.4 1.3 2.1 3.5 2.5 1.1U.K. 3.8 4.2 3.8 2.5 2.4 4.0

Wholesale trade Retail tradeMotor vehicle trade

Distributive trade sector accounts for large part of post-1995 productivity boom in U.S. (Triplett & Bosworth, 2004)Big part of EU-US productivity differential is traced to this sector (van Ark, Inklaar and McGuckin, 2002)

Combination of ICT and organisational innovation has transformed sector into “information industry” (McGuckin, Spiegelman and van Ark, 2004)

Page 3: Productivity Differentials in the U.S. and EU Distributive ... · Germany 1.4 -1.0 2.7 1.6 2.2 1.3 Netherlands 2.4 1.3 2.1 3.5 2.5 1.1 ... Changing business models in distributive

3

But is the EU-US productivity growth differential real or is it a

measurement issue?

Impact of “inside the box” quality adjustments for ICT on U.S. trade output measureExperimental estimates of double deflated real margin instead of real sales indexSources of productivity growth in trade: increase in service output or saving in intermediate inputs?

Page 4: Productivity Differentials in the U.S. and EU Distributive ... · Germany 1.4 -1.0 2.7 1.6 2.2 1.3 Netherlands 2.4 1.3 2.1 3.5 2.5 1.1 ... Changing business models in distributive

4

Main conclusions

There is a slight upward bias in U.S. trade output measures due to ICT but much more so in wholesale than in retail servicesDouble deflated estimates of real output margin do not change aggregate retail measure in U.S., but cause shifts across retail industriesDecline in sales price is main cause of productivity gain in trade, but saving in intermediate inputs add to U.S. advantageMeasurement issues suggest upward bias but U.S. advantage remains intact

Page 5: Productivity Differentials in the U.S. and EU Distributive ... · Germany 1.4 -1.0 2.7 1.6 2.2 1.3 Netherlands 2.4 1.3 2.1 3.5 2.5 1.1 ... Changing business models in distributive

5

Changing business models in distributive trade should focus attention

on determinants of gross margin ...

Labour

Intermediate inputs

Purchases of goods

sold

Sales of goods

Gross margin

Value added Capital

Page 6: Productivity Differentials in the U.S. and EU Distributive ... · Germany 1.4 -1.0 2.7 1.6 2.2 1.3 Netherlands 2.4 1.3 2.1 3.5 2.5 1.1 ... Changing business models in distributive

6

… in particular from perspective of real or volume growth

Variables Observed Imputed Values Imputed QuantitiesVariables

Sales Price (pS) pS

Sales Quantity (qS) qS = S - p S

Sales Value (S) S

Goods Purchases Value (C) C

Margin Quantity (qM) qM= qS

Margin Value (M) M = S - C

Intermediate Inputs Price (pII) pII

Intermediate Inputs Quantity (qII) qII = II - pII

Intermediate Inputs Value (II) II

Value Added Value (VA) VA = M - IIValue Added Quantity (qVA) QVA = w [qM - (1-w) qII]

Volume of trade services does not only depend on number of transactions (number of boxes) but also on quality of service provided, which is key for success of failure of industryThe quality of products (“inside the box”) must be distinguished from the quality of the retail service

Page 7: Productivity Differentials in the U.S. and EU Distributive ... · Germany 1.4 -1.0 2.7 1.6 2.2 1.3 Netherlands 2.4 1.3 2.1 3.5 2.5 1.1 ... Changing business models in distributive

7

EXERCISE 1: DOES ICT PRODUCT QUALITY AFFECT EU-US DIFFERENTIAL?

Not the share of ICT consumption but the price change of ICT products makes the

difference between US and Europe

Price change1995 2003 1995-2003

France 3.5 3.9 -5.8Germany 3.4 2.4 -3.5Netherlands 4.9 4.8 -4.8UK 5.2 5.6 -8.1US 4.2 4.1 -13.0

Share in goods

Share of audio-visual, photographic and information processing equipment in household consumption expenditure, and associated CPI

change

Page 8: Productivity Differentials in the U.S. and EU Distributive ... · Germany 1.4 -1.0 2.7 1.6 2.2 1.3 Netherlands 2.4 1.3 2.1 3.5 2.5 1.1 ... Changing business models in distributive

8

Excluding core industries that trade ICT products reduces U.S. retail sales volume marginally, but it is more substantial for

wholesale tradeContributions of Industries to Real Sales Growth in the U.S., 1987-2002

1995-2002

Industry name 1987-1995 1995-2002over 1987-

1995

total retail 3.1 4.6 1.5total retail excluding 4431 and 4541 2.7 3.8 1.2difference 0.4 0.8 0.4

total wholesale 4.1 4.2 0.1total wholesale excluding 4234 and 4236 2.9 2.7 -0.2difference 1.2 1.5 0.3

Real sales

4234 = Commercial equipment; 4236 = Electrical and electronic goods; 4431 = Electronics and appliance stores; 4541 = Electronic shopping and mail order houses

Page 9: Productivity Differentials in the U.S. and EU Distributive ... · Germany 1.4 -1.0 2.7 1.6 2.2 1.3 Netherlands 2.4 1.3 2.1 3.5 2.5 1.1 ... Changing business models in distributive

9

Exercise 1 on impact of ICT provides upper bound estimate of bias

Not the whole of ICT-related trade industries should be considered, but only products which use hedonic price indicesEuropean countries also partly use hedonic price indices or alternative quality adjusted method (e.g. high frequency matched model deflators for ICT)… but it may not be confined to ICT: on the whole, U.S. proceeds faster in applying quality adjusted price indices throughout

Page 10: Productivity Differentials in the U.S. and EU Distributive ... · Germany 1.4 -1.0 2.7 1.6 2.2 1.3 Netherlands 2.4 1.3 2.1 3.5 2.5 1.1 ... Changing business models in distributive

10

EXERCISE 2: DOES USE OF SALES INDEX GIVE BIAS TO U.S. ADVANTAGE?

Three methods:use of real sales index (common practice, but different degree of detail across countries)use of margin prices (experimental stage: BLS)use of double deflation of sales and purchased of goods sold

Double deflation cancels “inside the box” price change out --> better proxy of real margin in line with services providedProblems:

Data requirements high => purchase pricesSensitive to measurement error => margin is residual

Page 11: Productivity Differentials in the U.S. and EU Distributive ... · Germany 1.4 -1.0 2.7 1.6 2.2 1.3 Netherlands 2.4 1.3 2.1 3.5 2.5 1.1 ... Changing business models in distributive

11

Exercise 2a: aggregate double deflation for trade sector as a whole (retail + wholesale)

Calculation of change in purchase price:

( ) Di

IIi

ICi pwpwp &&& −+= 1

Calculation of change in implict margin price:

[ ]( )CMSM

M pvpv

p &&& −−= 11

Simplified flow of goods for retail goods

Wholesale and

retail trade

Final consumption

Imports

Domestic production

Page 12: Productivity Differentials in the U.S. and EU Distributive ... · Germany 1.4 -1.0 2.7 1.6 2.2 1.3 Netherlands 2.4 1.3 2.1 3.5 2.5 1.1 ... Changing business models in distributive

12

Implicit margin prices - based on double deflation - decline everywhere but with

differences across countriesSales, purchases and margin prices for household consumption goods

Sales Purchases of which: Marginprices prices Domestic Imports prices

France 1.9 0.7 0.9 -2.0 4.5Germany 1.7 1.3 1.3 -0.1 2.9Netherlands 1.0 0.4 0.5 -0.1 2.1UK 3.3 4.1 3.2 0.9 1.4US 2.2 2.2 1.7 0.5 2.0

Sales Purchases of which: Marginprices prices Domestic Imports prices

France 1.3 0.5 0.7 -0.2 2.9Germany 0.7 0.9 0.8 0.1 0.0Netherlands 1.3 1.0 1.0 0.0 1.7UK 0.4 0.3 0.7 -0.4 0.7US 0.6 0.9 1.1 -0.2 0.1

Note: Germany: 1991-1995 instead of 1987-1995

1987-1995

1995-2002

Page 13: Productivity Differentials in the U.S. and EU Distributive ... · Germany 1.4 -1.0 2.7 1.6 2.2 1.3 Netherlands 2.4 1.3 2.1 3.5 2.5 1.1 ... Changing business models in distributive

13

With double deflation US (and UK !) trade sectors retain growth advantage

over continental European countries & it upwardly biases Europe rather than U.S.

National Accounts

Double deflated

National Accounts

Double deflated

France 2.3 -0.1 2.3 -0.4Germany n.a. 1.1 n.a. 2.4Netherlands 2.9 2.7 2.6 2.1UK n.a. n.a. n.a. 6.9US 3.3 3.4 5.3 5.3Germany 1987-1995 refers to 1991-1995

1987-1995 1995-2002

Growth of retail trade gross output, National Accounts versus double deflated prices, 1995-2001

Page 14: Productivity Differentials in the U.S. and EU Distributive ... · Germany 1.4 -1.0 2.7 1.6 2.2 1.3 Netherlands 2.4 1.3 2.1 3.5 2.5 1.1 ... Changing business models in distributive

Exercise 2b: double deflation by retail category puts much higher demand on data

Imports

Domestic production

Consumption

Retail: Department

stores

$22bln$13bln $9bln

Wholesale

10%

24%

46%

20%

Page 15: Productivity Differentials in the U.S. and EU Distributive ... · Germany 1.4 -1.0 2.7 1.6 2.2 1.3 Netherlands 2.4 1.3 2.1 3.5 2.5 1.1 ... Changing business models in distributive

15

EXERCISE 3: IS U.S. OUTPUT GROWTH IN TRADE DRIVEN BY MORE OUTPUT OR

LESS INPUTS?Variables Observed Imputed Values Imputed Quantities

Variables

Sales Price (pS) pS

Sales Quantity (qS) qS = S - p S

Sales Value (S) S

Goods Purchases Value (C) C

Margin Quantity (qM) qM= qS

Margin Value (M) M = S - C

Intermediate Inputs Price (pII) pII

Intermediate Inputs Quantity (qII) qII = II - pII

Intermediate Inputs Value (II) II

Value Added Value (VA) VA = M - IIValue Added Quantity (qVA) QVA = w [qM - (1-w) qII]

Page 16: Productivity Differentials in the U.S. and EU Distributive ... · Germany 1.4 -1.0 2.7 1.6 2.2 1.3 Netherlands 2.4 1.3 2.1 3.5 2.5 1.1 ... Changing business models in distributive

A. Retail (annul average change, %)

1 Quantity of output x O/VA ratio 4.5 7.4 3.02 Difference between quantity of sales and output 0.3 0.3 0.03 Quantity of sales x O/VA ratio 4.8 7.7 2.94 Value of sales x O/VA ratio 8.3 7.7 -0.65 Price of sales x O/VA ratio 3.6 0.0 -3.6

6 Quantity of intermediate inputs x II/VA ratio 1.7 0.7 -0.97 Value of intermediate inputs x II/VA ratio 3.1 1.8 -1.38 Price of intermediate inputs x II/VA ratio 1.4 1.0 -0.4

9 (=1- 6) Quantity of value added 2.8 6.7 3.9

B. Wholesale

1 Quantity of output x O/VA ratio 6.3 5.8 -0.52 Difference between quantity of sales and output -0.2 0.1 0.33 Quantity of sales x O/VA ratio 6.0 5.8 -0.24 Value of sales x O/VA ratio 8.2 3.9 -4.35 Price of sales x O/VA ratio 2.1 -1.9 -4.1

6 Quantity of intermediate inputs x II/VA ratio 2.3 -0.6 -2.97 Value of intermediate inputs x II/VA ratio 3.4 -0.3 -3.78 Price of intermediate inputs x II/VA ratio 1.1 0.2 -0.8

9 (=1- 6) Quantity of value added 3.9 6.6 2.7

1987-1995 1995-20021995-2002

over 1987-95

1987-1995 1995-20021995-2002

over 1987-95

In U.S. retailing about a quarter of acceleration is due to saving in intermediate inputs

Page 17: Productivity Differentials in the U.S. and EU Distributive ... · Germany 1.4 -1.0 2.7 1.6 2.2 1.3 Netherlands 2.4 1.3 2.1 3.5 2.5 1.1 ... Changing business models in distributive

A. Retail (annul average change, %)

1 Quantity of output x O/VA ratio 4.5 7.4 3.02 Difference between quantity of sales and output 0.3 0.3 0.03 Quantity of sales x O/VA ratio 4.8 7.7 2.94 Value of sales x O/VA ratio 8.3 7.7 -0.65 Price of sales x O/VA ratio 3.6 0.0 -3.6

6 Quantity of intermediate inputs x II/VA ratio 1.7 0.7 -0.97 Value of intermediate inputs x II/VA ratio 3.1 1.8 -1.38 Price of intermediate inputs x II/VA ratio 1.4 1.0 -0.4

9 (=1- 6) Quantity of value added 2.8 6.7 3.9

B. Wholesale

1 Quantity of output x O/VA ratio 6.3 5.8 -0.52 Difference between quantity of sales and output -0.2 0.1 0.33 Quantity of sales x O/VA ratio 6.0 5.8 -0.24 Value of sales x O/VA ratio 8.2 3.9 -4.35 Price of sales x O/VA ratio 2.1 -1.9 -4.1

6 Quantity of intermediate inputs x II/VA ratio 2.3 -0.6 -2.97 Value of intermediate inputs x II/VA ratio 3.4 -0.3 -3.78 Price of intermediate inputs x II/VA ratio 1.1 0.2 -0.8

9 (=1- 6) Quantity of value added 3.9 6.6 2.7

1987-1995 1995-20021995-2002

over 1987-95

1987-1995 1995-20021995-2002

over 1987-95

In wholesale entire productivity gain in U.S. is due to saving of intermediate inputs

Page 18: Productivity Differentials in the U.S. and EU Distributive ... · Germany 1.4 -1.0 2.7 1.6 2.2 1.3 Netherlands 2.4 1.3 2.1 3.5 2.5 1.1 ... Changing business models in distributive

18

Are intermediate input savings real?

Saving on intermediate inputs might be sign of greater dominance of “lean retailing”, competitive pressures and new entries… but we need comparative evidence for European countries to assess the differencesIncome approach in U.S. might overstate value added growth and therefore understate growth in intermediate inputs

cross-check with census-based approach suggests such problem in wholesale but not in retail

Page 19: Productivity Differentials in the U.S. and EU Distributive ... · Germany 1.4 -1.0 2.7 1.6 2.2 1.3 Netherlands 2.4 1.3 2.1 3.5 2.5 1.1 ... Changing business models in distributive

19

Main conclusions

There is a slight upward bias in U.S. trade output measures due to ICT but much more so in wholesale than in retail servicesDouble deflated estimates of real output margin do not change aggregate retail measure in U.S., but cause shifts across retail industriesDecline in sales price is main cause of productivity gain in trade, but saving in intermediate inputs add to U.S. advantageMeasurement issues suggest upward bias but U.S. advantage remains intact