Price settings in online markets: Basic facts, international comparisons, and cross-border...

1

Price settings in online markets: Basic facts, international comparisons, and cross-border integration

Y. Gorodnichenko (UC Berkeley) and O. Talavera (Sheffield)NBER WP 20406

2

LAW OF ONE PRICE (LOOP)

• A very intuitive and simple concept• Basic ingredient/assumption in many models in international

economics• In the data:

– Large deviations from LOOP• Explanations:

– Frictions (distance, information, tariffs) – Sticky prices– The data are “poor” (e.g., compare price indexes)

3


• A very intuitive and simple concept• Basic ingredient/assumption in many models in international economics• In the data (country- or region-level):

– Large/Heterogeneous deviations from LOOP– Slow (if any) convergence to LOOP

• Explanations: – Frictions (distance, information, tariffs) – Sticky prices– The data are “poor” (e.g., compare price indexes)

4


• A very intuitive and simple concept• Basic ingredient/assumption in many models in international economics• In the data (country- or region-level):

– Large/Heterogeneous deviations from LOOP– Slow (if any) convergence to LOOP

• Explanations: – Frictions (distance, information, tariffs) – Sticky prices– The data are “poor” (e.g., compare price indexes)

5

A twist to LOOP: EX Pass-Through

• Macro data– Campa and Goldberg (2005): The United States has among the lowest

pass-through rates in the OECD, at approximately 25% in the short run and 40% over the longer run.

• Avoiding aggregation bias:– Imbs et al. (2005), Cruchini and Shintani (2008), Broda and Weinstein

(2008): the pass-through and the speed of price adjustment are higher when individual, narrowly-defined goods are considered

6

Determinants of PT and Speed of Adjustment

• Factors affecting the size of the pass-through. – Market structure, – market power (including adjustment of mark-ups), – tariffs, – presence of multinationals– importance of non-traded inputs for price stickiness of final goods

Menon (1996), Cardasz and Stollery (2001), Gaulier, Lahreche-Revil, and Mejean (2006), Goldberg and Hellerstein (2012), Mayoral and Gardea (forthcoming)

7

What do we do?

We use online prices for estimating:• Descriptive statistics• Exchange Rate PT, speed of Adjustment• Determinants of PT/SA

Why online prices?

8

ONLINE MARKETS• Highly integrated and fast growing markets, unlike brick and mortar retailers• Easy search for best prices• Price comparison for identical goods is easy across stores• Negligible physical cost of changing prices• Goods sold online are easy to ship and transactions costs are small• Geographical location of stores and consumers is largely irrelevant• Nearly impossible to discriminate consumers based on their location• Easy to collect data

We use these unique characteristics to re-examine one of the key puzzles using prices from U.S. and Canadian online sellers.

9

ONLINE MARKETS

The IMRG, the UK’s industry association for e-retail, estimated that total e-commerce sales for 2011 were £77bn (or approximately 10 percent of total retail sales in the UK economy). This represents an increase of 14 percent from 2010 while total retail sales increased by 1.7 percent in 2011. Importantly, not only have online sales risen during the past recession, but they have historically grown much faster. While US, UK and Japan are leaders in business to consumer e-commerce, China’s e-commerce market in particular has risen over 130% in 2011. There is also high growth of online retailing in Europe, with Poland (29%), Ireland (24%), Spain (24%), France (21%) having highest spending online.

11

ONLINE MARKETS• Highly integrated and fast growing markets• Easy search for best prices• Price comparison for identical goods is easy across stores• Negligible physical cost of changing prices• Goods sold online are easy to ship and transactions costs are small• Geographical location of stores and consumers is largely irrelevant• Nearly impossible to discriminate consumers based on their location• Easy to collect data


12

ONLINE MARKETS• Highly integrated and fast growing markets• Easy search for best prices• Price comparison for identical goods is easy across stores• Negligible physical cost of changing prices, minimal menu costs, flexible prices• Goods sold online are easy to ship and transactions costs are small• Geographical location of stores and consumers is largely irrelevant• Nearly impossible to discriminate consumers based on their location• Easy to collect data


13



14



15



16

ONLINE MARKETS

• Highly integrated and fast growing markets• Easy search for best prices• Price comparison for identical goods is easy across stores• Negligible physical cost of changing prices• Goods sold online are easy to ship and transactions costs are small• Geographical location of stores and consumers is largely irrelevant• Nearly impossible to discriminate consumers based on their location• Easy to collect data

17

Research on Online Prices

• Brynjolfsson and Smith (2000) compare online and conventional stores prices on books and CDs. – online prices are 9-16% lower than prices in regular stores– the changes in online prices are much smaller for online prices, – quotes of internet prices are quite dispersed even for precisely

defined goods.

18

Research on Online Prices

• Lunnemann and Wintr (2011) document stickiness of online prices in the U.S. and large European markets (Germany, France, Italy, and the U.K.). – internet prices change less often in the U.S. than in Europe (the

opposite is true for conventional stores). – Online prices are more flexible than their offline counterparts with

half of the spells ending within a month.

19

Dispersion in Online Markets

• Dramatic dispersion of prices in online markets because of – information frictions – sellers’ ability to discriminate consumers (e.g., based on what sellers

know about customers), – differences in advertisement (e.g., investment in building brand,

reputation, etc.).

Baye and Morgan 2001, Baye and Morgan 2004, Morgan, Orzen, and Sefton 2006, Baye and Morgan 2009

20

Boivin et al. (2012)

• Dynamics of online price differences across three book sellers: Amazon.com (and Amazon.ca), BN.com (Barnes & Noble website), and Chapters.ca. – price differentials (or relative quantities) for books react to

fluctuations in the relative price of foreign competition following exchange rate movement which is consistent with extensive market segmentation and pervasive violations of the law of one price.

21

Cavallo et al (2014)

• Prices for all products sold by Apple, IKEA, H&M, and Zara through their online retail stores in 85 countries. – the law of one price holds very well within currency unions, but does

not hold outside currency unions– good-level real exchange rates reflect differences in prices at the time

products are first introduced

22

Contribution to literature

• We use unique characteristics of online markets to re-examine LOOP using prices from U.S. and Canadian online sellers.

23

DATA COLLECTION• We use a popular price comparison website.• Every Saturday at midnight, a Tcl/python script has been triggered to collect webpages with price

information. • The script extracts

– good description, – unique manufacturing product number (MPN), – prices for each seller, – sellers' unique ids and reviews.

• Our price quotes are price before taxes and shipping/handling costs. • What we have:

– >140,000 goods – >14 million good-seller-week-country quotes. – 55 types of goods in four main categories: computers (20 types, e.g., laptops), electronics (14 types, e.g.,

GPS), software (11 type, e.g., computer games), and cameras (10 types, e.g., digital cameras).

26

Sample compositionUSA Canada

1 SeaBoom.com AgileElectronics

2 NextWarehouse.com MostlyDigital

3 Amazon.com B&H PhotoVideo4 TheNerds.net TigerDirect.ca

5 Amazon.com Marketplace OnHop

6 PCConnectionExpress Newegg.ca

7 CompSource Inc. Amazon.ca

8 MacConnection PC-Canada

9 Memory4Less.com DirectDialCanada

10 PROVANTAGE Cendirect.com

27

US vs CA

11.1

1.2

1.3

No

min

al E

xcha

ng

e R

ate

, U

SD

/CA

D

15

020

025

030

035

040

0P

rice

Oct Jan Apr Jul Oct Jan Apr Jul Oct2008 2009 2010

WD VelociRaptor 300GB Hard Drive, by seller, US

11.

11.

21.

3N

om

inal

Exc

hang

e R

ate,

US

D/C

AD

200

250

300

350

400

450

Pric

e

Oct Jan Apr Jul Oct Jan Apr Jul Oct2008 2009 2010

WD VelociRaptor 300GB Hard Drive, by seller, CA

28

DATA COLLECTION• We use a popular price comparison website.• Every Saturday at midnight, a Tcl/python script has been triggered to collect webpages with price

information. • The script extracts

– good description, – unique manufacturing product number (MPN), – prices for each seller, – sellers' unique ids and reviews.

• Our price quotes are price before taxes and shipping/handling costs. • What we have:

– >140,000 goods – >14 million good-seller-week-country quotes. – 55 types of goods in four main categories: computers (20 types, e.g., laptops), electronics (14 types, e.g.,

GPS), software (11 type, e.g., computer games), and cameras (10 types, e.g., digital cameras).

29

Start of data collection

.91

1.1

1.2

1.3

Ca

na

dia

n D

olla

rs t

o O

ne U

.S.

Do

llar,

Daily

01jan2008 01jan2009 01jan2010 01jan2011 01jan2012

30

DESCRIPTIVE STATISTICS

Canada USA Mean St.Dev Mean St.Dev (1) (2) (3) (4) Cross-sectional distribution of prices

St.dev. log(Price) 0.128 0.105 0.151 0.123 IQR log(Price) 0.095 0.097 0.160 0.151 Median log(Price) 5.077 1.488 4.979 1.471

Median duration of price spell (weeks) 2.974 9.615 4.112 9.335 Size of price changes

Median dlog(Price) -0.007 0.030 -0.005 0.031 Median abs(dlog(Price)) 0.041 0.058 0.042 0.048

Synchronization of price changes 0.299 0.213 0.190 0.121 Properties of sellers

Number of sellers 2.399 1.318 3.627 1.983 Stability 0.423 0.213 0.405 0.188

31








𝑆𝑦𝑛𝑐ℎ𝑟𝑜𝑛𝑖𝑧𝑎𝑡𝑖𝑜𝑛𝑖𝑡𝑐 = σ 𝟏൛𝑃𝑖𝑡𝑠𝑐 ≠ 𝑃𝑖,𝑡−1,𝑠𝑐ൟ𝑠∈𝒮𝑖𝑡𝑐 − 1σ 𝟏൛𝑃𝑖𝑡𝑠𝑐 ≠ missing ∩𝑃𝑖,𝑡−1,𝑠𝑐 ≠ missingൟ𝑠∈𝒮𝑖𝑡𝑐 − 1

32








33

INTERNATIONAL PRICE COMPARISONS

Mean St.Dev

AR1 OLS

N

(1) (2) (4) (5)

Panel A: Mean prices (net)

Relative exchange rate 0.083 0.236 0.923 1,709,011 Real exchange rate 0.052 0.231 0.917 1,709,011

Panel B: Mean prices (gross)

Relative exchange rate 0.107 0.205 0.929 830,169 Real exchange rate 0.064 0.205 0.928 830,169

Relative exchange rate ≡ log൫𝑃𝑖𝑡𝐶𝐴/𝑃𝑖𝑡𝑈𝑆൯ Real exchange rate ≡ log൫𝐸𝑋−1𝑡 × 𝑃𝑖𝑡𝐶𝐴/𝑃𝑖𝑡𝑈𝑆൯

34

PASS-THROUGH

Pass-through (PT): log൬𝑃𝑖𝑡𝐶𝐴𝑃𝑖𝑡𝑈𝑆൰= 𝛼𝐸𝑋𝑡 + 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑠+ 𝑒𝑟𝑟𝑜𝑟𝑖𝑡, (1)

The long-run pass-through and is similar to specifications estimated in Goldberg and Knetter (1997), Campa and Goldberg (2005), Goldberg and Hellerstein (2012).

The law of one price predicts that 𝛼 should be equal to one and larger values of 𝛼 correspond to smaller departures from the law of one prices.

35

SPEED OF PRICE ADJUSTMENT Speed of price adjustment: 𝑑log൬𝑃𝑖𝑡𝐶𝐴𝑃𝑖𝑡𝑈𝑆൰= 𝛽൬log൬𝑃𝑖,𝑡−1𝐶𝐴𝑃𝑖,𝑡−1𝑈𝑆 ൰− 𝛼𝐸𝑋𝑡−1൰ (2)

+𝜙1𝑑log൬𝑃𝑖,𝑡−1𝐶𝐴𝑃𝑖,𝑡−1𝑈𝑆 ൰+ 𝜆1𝑑𝐸𝑋𝑡−1 + 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑠+ 𝑒𝑟𝑟𝑜𝑟𝑖𝑡

It is set in the error-correction/cointegration form where 𝛽 quantifies how quickly the deviation from equilibrium is eliminated.

In Specification (2), equilibrium relationship between relative and the exchange rate are determined according to Specification (1).

While the equilibrium relationship nests the law of one price, it also allows deviations from the law of one price (i.e., α can be less than one).

36

Basic Results: Net Prices PT Speed (1) (2) Mean Price 0.698*** -0.158*** (0.080) (0.005) Median Price 0.698*** -0.170*** (0.082) (0.005) Minimum Price 0.677*** -0.169*** (0.042) (0.005) Within-seller 0.284*** -0.116*** (0.050) (0.013)

37

DECOMPOSITION OF PRICE ADJUSTMENT𝑀𝑖𝑐𝑡 = 𝛾𝑐 + 𝜓𝑐ቊlogቆ𝑃𝑖,𝑡−1𝐶𝐴𝑃𝑖,𝑡−1𝑈𝑆 ቇ− 𝛼ො��𝐸𝑋𝑡−1ቋ+ 𝜅𝑐1𝐸𝑋𝑡−1 + 𝜅𝑐2𝑀𝑖𝑐,𝑡−1 + 𝜆𝑖𝑐 + 𝑒𝑟𝑟𝑜𝑟𝑖𝑐𝑡

𝐸𝑞𝑢𝑖𝑙𝑖𝑏𝑖𝑟𝑢𝑚 𝑒𝑟𝑟𝑜𝑟≡ ൜log൬𝑃𝑖,𝑡−1𝐶𝐴𝑃𝑖,𝑡−1𝑈𝑆 ൰− 𝛼ො��𝐸𝑋𝑡−1ൠ> 0 ⇒ Canadian goods are expensive

38

DECOMPOSITION OF PRICE ADJUSTMENT𝑀𝑖𝑐𝑡 = 𝛾𝑐 + 𝜓𝑐ቊlogቆ𝑃𝑖,𝑡−1𝐶𝐴𝑃𝑖,𝑡−1𝑈𝑆 ቇ− 𝛼ො��𝐸𝑋𝑡−1ቋ+ 𝜅𝑐1𝐸𝑋𝑡−1 + 𝜅𝑐2𝑀𝑖𝑐,𝑡−1 + 𝜆𝑖𝑐 + 𝑒𝑟𝑟𝑜𝑟𝑖𝑐𝑡

𝐸𝑞𝑢𝑖𝑙𝑖𝑏𝑖𝑟𝑢𝑚 𝑒𝑟𝑟𝑜𝑟≡ ൜log൬𝑃𝑖,𝑡−1𝐶𝐴𝑃𝑖,𝑡−1𝑈𝑆 ൰− 𝛼ො��𝐸𝑋𝑡−1ൠ> 0 ⇒ Canadian goods are expensive

Moment, (mean prices)

CA US(1) (2)

Probability of price adjustment Any, -0.002 0.008Increase, -0.080*** 0.025***Decrease, 0.076*** -0.016***

Probability of exit -0.010 -0.006

39

GOOD-SPECIFIC PASS-THROUGH

Pass-through (PT): log൬𝑃𝑖𝑡𝐶𝐴𝑃𝑖𝑡𝑈𝑆൰= 𝛼𝐸𝑋𝑡 + 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑠+ 𝑒𝑟𝑟𝑜𝑟𝑖𝑡,

Speed of price adjustment: 𝑑log൬𝑃𝑖𝑡𝐶𝐴𝑃𝑖𝑡𝑈𝑆൰= 𝛽൬log൬𝑃𝑖,𝑡−1𝐶𝐴𝑃𝑖,𝑡−1𝑈𝑆 ൰− 𝛼𝐸𝑋𝑡−1൰+

𝜙1𝑑log൬𝑃𝑖,𝑡−1𝐶𝐴𝑃𝑖,𝑡−1𝑈𝑆 ൰+ 𝜆1𝑑𝐸𝑋𝑡−1 + 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑠+ 𝑒𝑟𝑟𝑜𝑟𝑖𝑡

Estimate 𝛼 and 𝛽 for each good separately and related variation of 𝛼 and 𝛽 across goods to “fundamentals” such returns to search, price stickiness, etc.

40

DETERMINANTS OF PASS-THROUGH AND SPEED OF PRICE ADJ.

A higher search intensity should put a larger pressure on price convergence across sellers and countries.

Regressors

Price measure: Min

Path-throughSpeed of

Adjustment(1) (2)

Log(Median Price) 0.546*** 0.007 (0.124) (0.010)Log(Median Price)2 -0.054*** -0.001 (0.011) (0.001)Freq. of price change 1.882*** -0.073***

(0.236) (0.026)Log(Sellers) 2.152*** -0.147** (0.400) (0.060)Log(Sellers)2 -0.666*** 0.044** (0.123) (0.018)Stability of Sellers 0.903** 0.337*** (0.398) (0.052)Synchronization -0.474* 0.039 (0.265) (0.028)Average Reputation 0.139** -0.005 (0.053) (0.005)Observations 11,713 11,713R2 0.11 0.13

41


Sticky prices could delay price adjustment and make it incomplete.

Imbs et al. 2005, Mayoral and Gardea forthcoming, Rogoff 1996, Takhtamanova 2008

Regressors

Price measure: Min


Adjustment(1) (2)



42


Pass-through and speed of price adjustment could be affected not only by the degree of price stickiness at the level of individual seller but also to what extent price setting is staggered.

Regressors

Price measure: Min


Adjustment(1) (2)



43


The number of sellers should be indicative of the degree of competition and thus can help discriminating price stickiness vs. market power

Regressors

Price measure: Min


Adjustment(1) (2)



44


Easy entry into a market and limited time-horizons for sellers (this limits the scope for collusion) are likely to eliminate arbitrage opportunities and mis-pricing of goods faster

Regressors

Price measure: Min


Adjustment(1) (2)



45

Concluding remarks

• Online retail is closer to a frictionless ideal market– More flexible prices

• Law of one price is a reasonable approximation– Mildly persistent price differentials– Pass-through is relatively high– Speed of price adjustment is high– All margins are working to eliminate price differentials

• Large variation across goods– This variation can be systematically related to “fundamentals”

• Soaring internet retail can bring the law of one price closer to reality

46

Concluding remarks





47

Concluding remarks





48

Concluding remarks





Price settings in online markets: Basic facts, international comparisons, and cross-border...

Documents

Transcript of Price settings in online markets: Basic facts, international comparisons, and cross-border...