International Trade
Session 3
- Comparative advantages: from traditional to newer versions
- Introduction: empirical methods
Vilém Semerák
[email protected]
IES, Fall 2024
https://dl1.cuni.cz/course/view.php?id=14117
Moodle password: Trading_2024$
Outline of This Session
• Course organization
• Theory of comparative advantage
• Ricardo in modern disguise
• Wages and exchange rates?
• More commodities?
• Seminar session – empirical methods
• RCA
• Indicator
• Some hints on handling the data
• Openness
• Concentration
• A brief test?
Reading/Listening for This Week
• KOM – chapter 3
• Helpman: Chapter 2 – introductory part + section 2.1
• Krugman: Ricardo’s difficult idea
Reading/Listening for the Next Week
• Helpman (2011): chapter 2 (section 2.2)
• KOM: chapter 5
• And try another podcast:
• Tradetalks: 191. Brazil’s trade opening and its toll on workers and crime:
https://tradetalkspodcast.com/podcast/191-brazils-trade-opening-and-its-toll-on-
workers-and-crime/
• Questions for the podcast:
• What happened with tariffs in Brazil? How can country’s comparative advantages be related to the
structure of the country’s tariffs?
• How did they test effects of the trade policy change on labour markets?
• So, what were the effects on labour markets? What did they find out about the timing of the effects?
• Which US city is mentioned in the podcast?
• What is the TAA mentioned in the podcast?
• How big is Brazilian informal sector? How was it influenced by the trade policy change?
• Does it seem that homicides were linked to changes in trade policy?
Quizzes and Teams
• Please try to form teams consisting of 3-4 members
• Does everybody have an idea about a team?
• Option to join a team will be open on Moodle
• Online quizzes
• The second one opened last week – with the deadline for
today
• The next quiz will be started by Sunday midnight
• Deadline: October 24th midnight
• On-site activity? How about today?
Comparative Advantages
Translated to Modern Economics
Assumptions of the Model
• Differences in technologies exist
• 1 factor of production (labour)
• Labour is homogenous, perfectly mobile within a country but
not internationally
• Perfect competition, all resources are fully used
• Balanced balance of trade (no credit, no investment!)
• Constant returns to scale
• 2 countries and 2 goods
• We neglect transportation costs and barriers to trade
• We do not assume the existence of money
Ricardian Example
Units of labour Units of labour
per 100 yards of per 100 litres of
cloth wine
England 100 120
Portugal 90 80
• How to find out the opportunity costs?
• England has no absolute advantage, but it still has a comparative
advantage in production of cloth
• Portugal has its comparative advantage in the production of wine
• But will the countries really trade?
The Same Situation: More Formal Treatment
• Some “facts” on “England”: 1
• Production functions: 𝑄𝑊 = ⋅ 𝐿𝑊
120
1
𝑄𝐶 = ⋅ 𝐿𝐶
100
• Total labour endowment: 3600 units (new info)
• Preferences:
• More general case: 𝑈 = 𝑢(𝐶𝐶 , 𝐶𝑊 )
• Simpler case (exercises): 𝑈 = min 𝐶𝐶 , 𝐶𝑊
• International price ratio: 100 yards of cloth/100
litres of wine
Wine England: PPF
30 𝐿𝐸 = 𝐿𝑊 + 𝐿𝐶
3600 = 120𝑄𝑊 + 100𝑄𝐶
25
PPF: 100
𝑄𝑊 = 30 − 𝑄𝐶
20 120
15
10
PPF
5 10 15 20 25 30 36 Cloth
Wine England: Optimal Specialization
30
25
20
15
10
PPF
5 10 15 20 25 30 36 Cloth
Wine
England: Trade & Autarky Equilibrium
CPF
What determines the slope?
30 How does structure of production change?
What are the welfare effects?
25
PPF: 100
𝑄𝑊 = 30 − 𝑄𝐶
120
20
CPF: 100
𝑄𝑊 = 36 − 𝑄𝐶
100
15
10
PPF
5 10 15 20 25 30 36 Cloth
Wine
England: Terms of Trade and Welfare
CPF
30
25
20
15
10
PPF
5 10 15 20 25 30 36 Cloth
Wine England (with Leontief Utility)
CPF
30
25
20
15
10
PPF
5 10 15 20 25 30 36 Cloth
Free Trade v. Autarky for England (1)
• We will use the advantages provided by the assumed
Leontief preferences:
• Autarky:
100
• PPF: 𝑄𝑊 = 30 − 𝑄𝐶
120
• Consumers’ optimum: 𝐶𝑊 = 𝐶𝐶
• Autarky implies: 𝐶𝑊 = 𝑄𝑊 and 𝐶𝐶 = 𝑄𝐶
• Autarky allocations:
• 𝐶𝑊 = 𝑄𝑊 ≅ 16.36
• 𝐶𝐶 = 𝑄𝐶 ≅ 16.36
• Utility (cardinal?) = 16.36
Free Trade v. Autarky for England (2)
• We will use the advantages provided by the assumed
Leontief preferences:
• Free trade: 100
• CPF: 𝐶𝑊 = 36 − 𝐶
100 𝐶
• Consumers’ optimum: 𝐶𝑊 = 𝐶𝐶
• Free trade allocations:
• QW =0, CW= 18 , Import = 18 (in physical units)
• QC =36, CC= 18, Export = 18
• Utility (cardinal?) = 18
Portugal:
• The same conversion:
• Total labour endowment: 3600 units
• Production functions: 1
𝑄𝑊 = ⋅ 𝐿𝑊
80
1
𝑄𝐶 = ⋅ 𝐿𝐶
90
• Preferences: 𝑈 = min 𝐶𝐶 , 𝐶𝑊
• International price ratio: 100 yards of cloth/100
litres of wine
Wine Portugal
45
Again: use the data from the table + assume
a particular level of labor endowment (here 3600 units)
CPF
PPF
40 Cloth
Identification of Comparative
Advantage
• Note that the pattern of comparative advantage is
revealed by:
• Ratios of labour requirements (1/labour productivities)
• Autarky price ratios (relative prices)
• Slope of the PPFs (marginal rate of transformation)
• And – as we cam simply demonstrate – this is due to the
fact that these are mathematically identical
Equilibrium Pattern of Trade
In a global equilibrium the excess demands and
supplies of the two countries must exactly fit
together:
(C A
X − Q XA ) = − (C XB − Q XB ) and (C
Y
A
− QYA ) = − (CYB − QYB )
- A fairly restrictive condition especially if the
hypothetical model world only consists of two
countries
World’s Relative Supply and Demand
Relative price
of wine
PW
PC
aWE RS
aCE
RD2
aWP
RD1
aCP
QW
QC
LP aWP Relative quantity of wine
LE aCE
World’s Relative Supply and Demand: Our Case
Relative price
of wine
PW RD
PC
120
RS
100 English autarky price
80
Portuguese autarky price
90
QW
45 Relative quantity of wine
36 QC
Questions:
• What happens if England gets bigger (Portugal gets
smaller)?
World’s Relative Supply and Demand: Our Case
Relative price
of wine
PW RD
PC
120
RS
100 English autarky price
80
Portuguese autarky price
90
45
QW
45 Relative quantity of wine
72 36 QC
Questions (2):
• What happens if consumers preferences change to
something like this:
𝐶𝑊
𝑈 = min 𝐶𝐶 ,
2
World’s Relative Supply and Demand: Our Case
Relative price
of wine
PW RD RD’
PC
120
RS
100 English autarky price
80
Portuguese autarky price
90
QW
45 Relative quantity of wine
36 QC
Ricardian Model - Extensions:
(i) Wages and Comparative Advantages
(ii) Ricardian model with many goods
Simple Extension: Wages and
Exchange Rates
• We must be sure that comparative advantage not only
identifies opportunities for improvement, but that
these will be used
• Comparative advantage (labour requirements) and absolute
advantage (in costs of production)
• Additional motive: In the real world, countries seem to
have the ability to manipulate levels of wages and
exchange rates. What happens then?
• On the margin:
• This extension was used by Samuelson in his 1964
explanation of the famous Balassa-Samuelson effect
Back to the old Ricardian Example…
Units of labour Units of labour
per 100 yards of per 100 litres of
cloth wine
𝐸𝑛𝑔𝑙𝑎𝑛𝑑 𝐸𝑛𝑔𝑙𝑎𝑛𝑑
England 𝑎𝑐𝑙𝑜𝑡ℎ 𝑎𝑤𝑖𝑛𝑒
𝑃𝑜𝑟𝑡𝑢𝑔𝑎𝑙 𝑃𝑜𝑟𝑡𝑢𝑔𝑎𝑙
Portugal 𝑎𝑐𝑙𝑜𝑡ℎ 𝑎𝑤𝑖𝑛𝑒
Trade, Wages, Nominal Prices and
Exchange Rates
• Assume that England and Portugal now use money
(pounds and escudos) instead of barter exchange
• There is some level of nominal wages in each of the
countries: wE, wP
• There is also some exchange rate that gives price of
escudos in pounds (E)
• Nominal prices in England (in pounds):
𝐸𝑛𝑔𝑙𝑎𝑛𝑑 𝐸𝑛𝑔𝑙𝑎𝑛𝑑
𝑝𝑐𝑙𝑜𝑡ℎ = 𝑎𝑐𝑙𝑜𝑡ℎ ⋅ 𝑤𝐸
𝐸𝑛𝑔𝑙𝑎𝑛𝑑 𝐸𝑛𝑔𝑙𝑎𝑛𝑑
𝑝𝑤𝑖𝑛𝑒 = 𝑎𝑤𝑖𝑛𝑒 ⋅ 𝑤𝐸
• Nominal prices in Portugal (in pounds!):
𝑃𝑜𝑟𝑡 𝑃𝑜𝑟𝑡
𝑝𝑐𝑙𝑜𝑡ℎ = 𝑎𝑐𝑙𝑜𝑡ℎ ⋅ 𝑤𝑃 ⋅ 𝐸 𝑃𝑜𝑟𝑡
𝑝𝑤𝑖𝑛𝑒 𝑃𝑜𝑟𝑡
= 𝑎𝑤𝑖𝑛𝑒 ⋅ 𝑤𝑃 ⋅ 𝐸
• If England exports cloth, it should hold that:
𝐸𝑛𝑔𝑙𝑎𝑛𝑑 𝑃𝑜𝑟𝑡
𝑎𝑐𝑙𝑜𝑡ℎ ⋅ 𝑤 𝐸 ≤ 𝑎𝑐𝑙𝑜𝑡ℎ ⋅ 𝑤𝑃 ⋅ 𝐸
• If Portugal exports wine, it should hold that:
𝐸𝑛𝑔𝑙𝑎𝑛𝑑 𝑃𝑜𝑟𝑡
𝑎𝑤𝑖𝑛𝑒 ⋅ 𝑤 𝐸 ≥ 𝑎𝑤𝑖𝑛𝑒 ⋅ 𝑤𝑃 ⋅ 𝐸
• Consequently:
𝐸𝑛𝑔𝑙𝑎𝑛𝑑
• Explain why? 𝑎 𝐸𝑛𝑔𝑙𝑎𝑛𝑑 𝑤 ⋅𝐸𝑃 𝑎𝑤𝑖𝑛𝑒
𝑐𝑙𝑜𝑡ℎ
𝑃𝑜𝑟𝑡 ≤ 𝐸
≤ 𝑃𝑜𝑟𝑡
𝑎𝑐𝑙𝑜𝑡ℎ 𝑤 𝑎𝑤𝑖𝑛𝑒
A Few Questions
• So what happens if English labour unions push
through a new law which mandates a minimum
wage at the Portuguese level (wE=wPE)?
• When will English wage catch up with the
Portuguese wage?
Ricardo with Many (n) Goods
Ricardo with Many Goods (k>2)
Home unit Foreign unit
Relative
labour labour
Good productivity
requirements requirements
ai*/ai
ai ai*
Apples 1 10 10
Bananas 5 40 8
Caviar 3 12 4
Dates 6 12 2
Enchiladas 12 9 0.75
Ricardo with Many Goods (k>2)
Relative Home price Foreign
Good ai ai* productivity for price for
ai*/ai w=3 w* = 1
Apples 1 10 10 3 10
Banan
5 40 8 15 40
as
Caviar 3 12 4 9 12
Dates 6 12 2 18 12
Enchil
12 9 0.75 36 9
adas
Ricardo with Many Goods (k>2)
Relative Home price Foreign
Good ai ai* productivity for price for
ai*/ai w=3 w* = 1
Apples 1 10 10 3 10
Banan
5 40 8 15 40
as
Caviar 3 12 4 9 12
Dates 6 12 2 18 12
Enchila
12 9 0.75 36 9
das
Relative Wages and Trade
𝑎1∗ 𝑎2∗ 𝑤 𝑎𝑛∗
< <⋯< ∗<⋯<
𝑎1 𝑎2 𝑤 𝑎𝑛
• Relative wages (after conversion to comparable
currency) determine what the countries should
import and export.
• However, relative wages are at the same time
determined by labour requirements, structure
of demand and size of the economies.
World’s Relative Supply and Demand
Relative price
of wine
PW
PC
aWE RS
aCE
RD2
aWP
RD1
aCP
QW
QC
LP aWP Relative quantity of wine
LE aCE
Source: KOM, chapter 3
A Few Thought Experiments
• What happens if:
• Our country’s labour force shrinks (at least in relative terms)?
• What happens with our specialization (pattern of trade), welfare?
• What happens if there is migration from the rest of the world to our
economy?
• Note: if you want to play more with this extended version of
the model, there is a nice paper:
• Dornbusch, Fischer & Samuelson (1977): Comparative Advantage,
Trade, and Payments in a Ricardian Model with a Continuum of
Goods. AER
• If interested, we can discuss this version of the model during
consultations or in one of the next seminar sessions.
Empirical Issues
Empirical Tests
• So does the Ricardian model fit the reality?
• Kind of…
• Productivity of labour really seems to matter (for wages
and at least some patterns)
Productivity of Labour and the Pattern of Trade
Chart from Krugman, based on Balassa (1963)
FIGURE 4-9 Labor Productivity and Wages
Shown here are estimated
labor productivities across
countries, and their
wages, relative to the
United States in 1990.
Notice that the labor and
wages were highly
correlated across
countries: the points
roughly line up along the
45-degree line.
Source: Feenstra & Taylor
Ricardian Model:
Conclusions, Weaker & Stronger
Features
Ricardian Model: Conclusions
• Trade is caused by differences in technologies (relative
productivities of labour)
• Corresponds with some patterns of trade data
• Profit and utility maximization of rational agents guarantee
that trade leads to higher (or at least equal) welfare for
everyone
• Welfare improvements caused by specialization
• Relative wages also dictated by productivities, low wages of
poor countries are not social dumping but necessity
• Problems:
• Long-run perspective (short-run costs and imperfections
neglected) – with some inherent inconsistencies
• No real dynamics (how about e.g. interaction between trade
specialization and e.g. human capital accumulation?)
Weaker Points of Traditional CA
• Crucial role of mobility (and homogeneity) of labour
• There are some omissions/simplifications:
• It is static – what if there can be some interaction between
current specialization and future development of the
economies?
• Trade is based on differences in productivity (technologies). But
what about countries that are very similar but still trade? Can
there be other reasons for trade?
• What happens if we introduce imperfect competition and IRS
instead of perfect competition and CRS?
• Can we analyze how trade influences income distribution?
• Countries often seem to trade similar commodities (intra-
industry trade). Is this consistent with the Ricardian model?
Interesting Newer Papers
• Eaton & Kortum (2012): Putting Ricardo to Work (JEP)
• Provided on moodle (but the whole text is not compulsory)
• A very recent use:
• Ossa (2024) - Rethinking Ricardo: Environmental Comparative
Advantage and the Environmental Gains from Trade
• https://www.oenb.at/dam/jcr:38c59690-6e6b-45b2-859d-
e296c810f0f4/Greening-Ricardo.pdf
• Ricardo and intra-industry trade
• Davis (1995): Intra-industry trade: A Heckscher-Ohlin-Ricardo
approach
• Ricardo and economies of scale
• Gomory (1994): A Ricardo Model with Economies of Scale
• Ricardo and geographical features
• Eaton & Kortum (2002): Technology, Geography, and Trade
Revealed Comparative Advantages
Revealed Comparative Advantages
• Simple idea – if we have a comparative advantage, we
should be exporting the product
• So: let us find out which products we are exporting than an
average country
• Basic form: i … sector
𝑋𝑖𝑗 Τσ𝑖 𝑋𝑖𝑗 j … country
𝑅𝐶𝐴𝑖𝑗 =
• Balassa (1965): σ𝑗 𝑋𝑖𝑗 ൗσ𝑖 σ𝑗 𝑋𝑖𝑗
Benchmark – export
patterns
of the world or of e.g.
OECD (or EU 27)
• Normalized version:
𝑅𝐶𝐴𝑖𝑘 − 1
𝑁𝑅𝐶𝐴𝑖𝑘 = Values -1 ... +1
𝑅𝐶𝐴𝑖𝑘 +1
• Possible issues:
• Trade deficits (surpluses)
• Real importance of sectors (their share in total production)
• Two-way trade
RCA Index: Some Practical Issues
• Calculations:
• Excel/LibreOffice: pivot tables useful for sorting the data
• It can be done elegantly in Stata or R (examples will be
shared online).
• What would you use the RCA index for?
• Assume that I want to analyze how Czech exports to
Brazil will be influenced by the new EU-Mercosur trade
agreement
• Of course: other interesting (and often more
sophisticated indicators) exist, we will discuss some
of them in following weeks
A Brief Empirical Intro
Openness of Economies
How do we Compare Openness?
• Why this can be useful?
• We can get a quick idea on how sensitive to external
shocks a country can be
• Typical features:
• Smaller and richer countries are more open
• Main types of indicators:
• Indicators based on comparing “trade” with output
• Indicators based on the comparison of “trade” with
population
• E.g. Exports on capita
The Main Option: trade/Y
• Main versions:
• Exports/GDP (%)
• (Exports + Imports)/GDP (%)
• (Exports + Imports)/2GDP (%)
• Advantages: logical, easy to get
• Possible issues:
• Many versions, be careful which one you use
• Which measure of output to use?
• GDP, GNP,…
• GDP in PPP?
• GDP in PPP can be better, because poorer countries have currencies undervalued
to parity. Such countries would appear more open if we use GDP without the PPP
adjustment
• Can you explain why?
• Where do I get the data?
• WTO trade profiles
• World Bank World Development Indicators
Data from WTO Trade Profiles
Trade Trade per capita
(% of GDP, 2018-2020) US$ (2018-2020)
Armenia 43.2 1865
Czechia 71.4 16570
Georgia 53.9 2452
Kazakhstan 31 2913
Kyrgyz Rep. 49.4 624
Ukraine 44.6 1556
Hong Kong 184.5 87918
Source: https://www.wto.org/english/res_e/statis_e/trade_profiles_list_e.htm
• Specification:
• Trade to GDP ratio is estimated as an economy’s trade of goods and
commercial services (average of exports and imports, balance of payments
basis) divided by GDP (nominal)
• Why can it be useful to use average value (over several years)?
Specific Issues
• How can exports or imports (or both) of a country be bigger
than its GDP?
• Examples: Hong Kong, Singapore
• A typical cause “Entrepôt trade”
• The term entrepôt, also called a transshipment port and historically
referred to as a port city, is a trading post, port, city, or warehouse where
merchandise may be imported, stored, or traded before re-export, with no
additional processing taking place and with no customs duties imposed.
• Source: Investopedia
• A less likely cause (or a factor with minor contribution):
• While GDP measures output in a given year, trade data might include goods
produced in previous years
• Example: some indicators of trade between Czechia and Germany are
influenced by the presence of a significant flow of second-hand cars from
Germany….
• A very specific shock: e.g. German reunification and East Germany
How about the USA, EU, China?
Trade Trade per capita
(% of GDP, 2019-2021) US$ (2019-2021)
USA 12.6 8 287
China 18.2 2 007
Russia 24.7 2 789
EU 21.3 7 677
Czechia 69.7 16 972
Source: https://www.wto.org/english/res_e/statis_e/trade_profiles_list_e.htm
Do there seem to be any regularities?
Openness: A Simple Task to Try
• Let us try to analyze correlations between
openness and the level of development (GDP per
capita, PPP) and the size of countries (population)
• Data provided on Moodle
• In Excel – you can try to test the correlations or carry out a
simple regression directly in Excel or you can use e.g. Jamovi
• Jamovi (an open an easy-to-use econometric software):
https://www.jamovi.org/
• But also in another form – as Stata do-file and R script which
get the data directly and carry out simple analysis
Concentration/Diversification of
Trade Flows
Concentration Indices
• Why do we use them?
1. To measure the relative choice/market power of
suppliers/buyers
• Is there a difference between having 2 potential suppliers or 100?
• Is there a difference between being the only exporter in the whole
world, or one of many?
2. To measure exposure
• Does our economy depend on a relatively narrow set of products /
narrow group of trade partners?
• We are interested in concentration in several forms….
• Over products
• Over trade partners
• Or even in bilateral relations (how many types of products
with a particular partner)
Analysis of Export Concentration
(Commodities/Export Markets)
• How many products (product lines) (in some reasonable
minimal volume) / to how many countries do we
export?
• n (the number of categories, or the number of countries)
• Other simple option:
• The share of top 5 product categories (top 5 trade partners)
in total exports of the country?
• More complex indicators:
• Hirschman index (Herfindahl index/Hirschman-Herfindahl
index)
• And some other: Gini, Theil coefficient (not discussed now),
entropy index
Herfidahl (Hirschman, HH) Index of
Concentration
• Basic version: ℎ𝑖 = 𝑖 2
𝑠𝑘
𝑘
• ski = share of sector k in exports/imports of country i
• K … total number of product categories
• Values from 1/K (perfect dispersion) to 1 (complete concentration)
• Normalized version (with values from 0 to 1):
ℎ𝑖 − 1/𝐾
𝑛ℎ𝑖 =
1 − 1/𝐾
• The same formula can be used to calculate concentration
across countries (or of products within country pairs)
Normalized HH index
HS4 data, 4 digits, 2017
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0
CHE CHN CZE DEU JPN MMR RUS USA
Tests of the Relationship between
GDP p.c. and Diversification
Source: Cadot et al.
Diversification Comparisons
• Problem:
• It seems to be related to the level of development
• Another possible factor: natural resources
• We should account for this when testing whether
our country has more or less diversified trade than
other countries
References and Sources of Data
• Krugman: Ricardo’s difficult idea
• KOM – chapter 3
• Feenstra & Taylor – chapters 2-3
• A. Smith – Wealth of Nations
• A Practical Guide to Trade Policy Analysis – Chapter 1
(details on RCA indices)
