December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin...

56
The Visible Hand December 2014, Volume XXIII, Number I Ethan Poskanzer Syracuse University The Birth of the Silicon Wadi: The Effect of Post-Soviet Migration on Israeli Output Mix Sui Zhang Purdue University The Impact of VAT Hike on the Japanese Economy Nahiomy Alvarez Williams College When National and Local Policies Clash: How Seattle’s Increase in Minimum Wage Could Affect EITC Eligibility Pratyusha Mukherjee Princeton University Crouching Tiger, Rising Dragon: The Effect of China’s Oil Consumption on Singapore’s Economy Yaoyuan Liu, Haoran Chang, Tucker Phillips University of Wisconsin - Madison Competition in the Art Auction Industry

Transcript of December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin...

Page 1: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 1

Volume xxiii. issue i.

The Visible HandDecember 2014, Volume XXIII, Number I

Ethan PoskanzerSyracuse University The Birth of the Silicon Wadi: The Effect of Post-Soviet Migration on Israeli Output Mix

Sui Zhang Purdue University The Impact of VAT Hike on the Japanese Economy

Nahiomy Alvarez Williams College When National and Local Policies Clash: How Seattle’s Increase in Minimum Wage Could Affect EITC Eligibility

Pratyusha MukherjeePrinceton University Crouching Tiger, Rising Dragon: The Effect of China’s Oil Consumption on Singapore’s Economy

Yaoyuan Liu, Haoran Chang, Tucker Phillips University of Wisconsin - Madison Competition in the Art Auction Industry

Page 2: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

2 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

The Visible HandISSN: 1559-8802

Editor-in-Chief:Jacob Miller

Executive Board: Kristina Hurley

Sanjana KanthimathinathanNabeel MominManu Sharma

Nivedita Kutty Vatsa

Editors and Referees:Charles Anyamene

Monica CaiJesse ClurmanHui Dong HeRichard He

Kristina HurleyRishu Jain

Sanjana KanthimathinathanDan Liu

Jacob MillerAnthony Mohammed

James O’ConnorPriyanka Panigrahi

Jason SzeNivedita Kutty Vatsa

Maggie Wong

© 2014 Economics Society at Cornell. All Rights Re-served.

The opinions expressed herein and the format of citations are those of the authors and do not represent the view or

the endorsement of Cornell University and its Economics Society.

The Visible Hand thanks:

Jennifer P. Wissink, Senior Lecturer and Fac-ulty Advisor, for her valuable guidance and kind

supervisionThe Student Assembly Finance Commissionfor their generous continued financial support.

Table of Contents

3 Editorial Jacob Miller4 The Birth of the Silicon Wadi: The Effect of Post-Soviet Migration on Israeli Output Mix Ethan Poskanzer14 The Impact of VAT Hike on the Japanese Economy Sui Zhang26 When National and Local Policies Clash: How Seattle’s Increase in Minimum Wage Could Affect EITC Eligibility Nahiony Alvarez35 Crouching Tiger, Rising Dragon: The Effect of China’s Oil Consumption on Singapore’s Economy Pratyusha Mukherjee47 Competition in the Art Auction Industry Yaoyuan Lin, Haoran Chang, Tucker Phillips

Issues of The Visible Hand are archived at http://rso.cornell.edu/ces/publications.html

The Visible Hand is published each fall and spring with complimentary copies available around the

Cornell campus.We welcome your letters to the editor and comments! Please direct correspondence to the Editor-in-Chief at

Cornell Economics SocietyDepartment of Economics

Uris Hall, 4th FloorCornell UniversityIthaca, NY 14853

or [email protected]

Page 3: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 3

Volume xxiii. issue i.

Since economic recovery from the 2008 recession began in earnest, arguably the most popular topic in Economics has become wealth inequality in the United States. The unexpected success of the Occupy Wall Street movement first captured national attention in a way that shifted discussion away from the aftermath of the recent financial crisis to inequality on a national level. Though it is now clear that national wealth inequality has been rising since at least the 1980’s, Occupy Wall Street popularized the 1% vs. 99% conflict, setting the groundwork for future class conflicts in a nation that has long shied away from acknowledging class despite an ever stratifying society. Two recent works by prominent Economists managed to catch the public’s attention and facilitate continued discussion of inequality in the national discourse: Joseph Stiglitz’s The Price of Inequality in 2012 and Thomas Piketty’s Capital in the Twenty-First Century in 2014. Piketty’s work became a hit commercially and critically this year. It has become clear that today’s dauntingly uneven distribution of wealth across society is now a threat to the health of our very economic system. While high end income and wealth have been growing in a rapidly accelerating rate, income and wealth for everybody else have been stagnant or worse for decades. Concealing the stagnation of society’s middle and lower classes have been the entrance of women into the labor force, perilously high levels of public and private debt, and until the financial crisis the soaring value of homes. The United States’ next generation of leaders is now burdened with staggering amounts of college loans starting the morning after graduation. Our society is engaged in a massive transfer of wealth from the future to sustain consumption in the present. Such structural failures of the American political and economic systems have cast a shadow of doubt over the future and leave those who understand the consequences nervous. Peeling off the veneer and fixing the core of our economic and political systems must be the ultimate priority of our generation if we hope to preserve the Western society that countless

generations before have built. This semester’s journal features one of the broadest ranges of topics in recent memory. The excellent research will take readers to Israel, Singapore, Japan, New York, and Seattle, where the theme of inequality reemerges where a city will become a real world experiment in minimum wages. Thank you to all the authors who submitted articles this semester. It is always a privilege to read the products of hours of hard work and deep thought. Everyone from the Visible Hand appreciates the willing support from the Cornell Economics Society and Professor Wissink. We hope you enjoy reading the following papers as much as we did.

Jacob MillerEditor-in-Chief

Page 4: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

4 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

I. Introduction

Prior to the collapse of the Soviet Union, Rus-sian citizens were subject to strict emigration re-strictions. Following the fall of the USSR in the early 1990’s, large amounts of Jews began to mi-grate from Soviet Union to Israel. The population flow was one of the largest migrations relative to the population of the destination country in modern his-tory. The migration was unique, as the immigrants were significantly more likely to be educated and skilled than the native population. Additionally, evi-dence shows that after a short lag period, the major-ity of immigrants found skill appropriate labor. This paper will determine whether the influx of high skilled labor affected Israeli production pat-terns, and in what ways. This will be done through an analysis of output and export patterns based on skill intensity. The data for this analysis will primar-ily be from the OECD’s STAN database. Addition-ally, total employment levels by sector, as well as productivity per worker in the high tech sector will be examined to shed further light on the migration’s effect on the evolution of Israeli production patterns and the behavior of the Israeli technology sector.

Data for this analysis will come from STAN and the Israeli Central Bureau of Statistics. This paper will also attempt to rule out potential confounding expla-nations for the observed changes in production pat-terns. Finally, the paper will conclude with policy implications, and a discussion of how the results of this study can help countries in the contemporary world.

II. Background Information

1. Context of Migration to Israel The creation of the State of Israel in 1948 marks the most important event in the history of Jewish migration patterns. As a result of the Law of Return, passed in 1950, any officially Jewish person may become a citizen of the State of Israel. Traditionally, this means any child of a Jewish Mother or those who have undergone a formal conversion. Follow-ing the passage of the Law of Return, the country began to absorb large numbers of Jews from the Balkans, Eastern Europe, the Middle East and North Africa. This made Israel the second largest Jewish population center on Earth, following the United States. Notably, a minimal amount of the very large

Ethan Poskanzer Syracuse University

The Birth of the Silicon Wadi: The Effect of Post Soviet Migration on Israeli Output Mix

Technology is a crucial piece of the Israeli economy, making up 53.38% of overall manufacturing output. This has been true since Israel experienced rapid growth in high skill production during the late 1990‘s. This paper will focus on the effect of migration from the Former Soviet Union to Israel, and show that this was a driver of technology growth in Israel. Since the birth of the State of Israel in 1947, Jewish migrants from across the world have moved to the young nation. Migration flows peaked in the early 1990’s, follow-ing the fall of the Soviet Union. Soviet migrants were far more likely to be educated and skilled than Israeli natives at the time of the migration. Evidence shows that these migrants were likely to work in the same

EditorialSince its founding in 1947, Israel has been a popular destination for Jewish immigrants from around the world; particularly those from the Soviet Union after its disintegration in the early 1990s. In the period that followed, Israel witnessed tremendous economic advancement, primarily in the skill-intensive sectors of the economy. Ethan Poskanzer conducts a thorough analysis of the economic implications of this immigra-tion shock. His paper examines the changes in the compositions of the Israeli labor force, economic output, and, export. This discussion is particularly relevant as it comes at a time when immigration and its ramifica-tions are a contentious topic of public debate.

Nivedita Kutty Vatsa

Page 5: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 5

Volume xxiii. issue i.

Jewish population in the Soviet Union migrated upon the foundation of Israel. In 1970, the nations of the Former Soviet Union represented the third-larg-est population of Jews in the world, after Israel and North America. The Soviet Union contained 14% of the world’s Jews. Despite this, only about 1% of all immigrants to Israel during the first decade follow-ing the passage of the Law of Return came from the Soviet Union1. According to DellaPergola, this is largely due to political factors and limits placed on emigration by the Communist Party2. The trend of low Russian-Israeli migration was reversed following the collapse of the Soviet Union in the early 1990’s. In 1990, when the demise of the Soviet Union was imminent, emigration restrictions became ineffective. Shortly after, a massive amount of Jews migrated from the Soviet Union to Israel. This movement is known as the Post-Soviet Aliyah. From 1970 to 1988, 165,000 Jews emigrated from the USSR to Israel3. In the 1980’s, only about one thousand Soviets migrated to Israel monthly. In the 1990’s, close to one million Jews migrated from the nations of the Former USSR to Israel. In 1990 and 1991 alone, over 330,000 Jews migrated from the Former Soviet Union to Israel. At the peak of the wave, 36,000 Soviets migrated to Israel in a single month. The Soviet Union accounted for 91% of mi-gration to Israel during this time period. As a result of the Post-Soviet Aliyah, the Israeli population increased by 7% in just two years, and by 15% by the mid 1990s. During this time period, the Israeli work force grew by 8%4. The annual immigration flow is shown in Figure 1. The Post Soviet Aliyah represents a massive immigration and labor shock, one unparalleled in recent Western history. The im-migration shock caused the population of Israel to grow by 4% in 1990, and by an average of 1.7% annually over a seven year period. In comparison, the annual immigration rate in the United States peaked at about 1% of the population annually, and currently stands at 0.35% per year5.

2. Discussion of Education Levels of Immigrants v. Natives The Post Soviet Aliyah was a unique labor shock due to the unusually high education and skill levels of the immigrants. Traditionally, large migra-tions consist of relatively unskilled workers seek-ing economic opportunity. However, migrants from the Soviet Union to Israel were moving for other reasons, and the group as a whole had an unusually high education and skill level. According to Cohen-Goldner and Paserman, of those of age to enter the work force, 48.84% of males, and 48.44% of fe-

male immigrants from the Soviet Union to Russia between 1989 to 1994 held college degrees. Dur-ing this period, just 23% of male Israeli natives and 21.84% of female natives had completed a university education. Immigrants were also much more likely to have completed some college, while natives were much more likely to have less than a high school education. 32.4% of male natives and 29.26% of female native Israelis had less than a high school education, while just 8.33% of male and 7.77% of female immigrants had not completed a high school level education4. These statistics can be seen in Figures 2 and 3. In addition to being more educated, Russian immigrants were more likely to be employed in skill-intensive production sectors. According to the Israeli Labor Force Survey and Cohen-Goldner/Paserman, male immigrants who moved to Israel between1989 and 1993 were more likely to be em-ployed as academic & scientific professionals, ser-vice workers and skilled industry workers than na-tives. Statistics for females immigrants followed a similar trend. This survey excluded ultra-orthodox Jews, who generally do not join the labor force, im-migrants under the age of 25 and those who reported at least 30 years of schooling. Comparatively, na-tives were most likely to be employed as sales and clerical workers. It is notable that despite being more likely to be educated, initially immigrants were more likely to be employed as purely unskilled workers. Accord-ing to Cohen-Goldner & Paserman, this shows that the immigration shock created difficulties for immi-grants seeking skill appropriate labor. According to Eckstein & Weiss, the language barrier and issues adapting to new labor market institutions caused the delay for immigrants seeking skill appropriate labor. It is important to note that the percentage of natives working in skilled labor shifted following the im-migration shock4. This shows that the immigration shock affected native behavior, and was substantial enough to affect the labor market. It is likely that the inflow of skilled labor contributed to the growth of high skilled output and exports in Israel during the period following the migration.

III. Literature Review

The Post-Soviet Aliyah is a very unique migra-tion due to the volume of highly educated migrants and the small size of the Israeli economy. Accord-ingly, much literature has been written on the migra-tion’s economic implications. One of these, Rebecca Friedberg’s The Impact of Mass Migration on the

Page 6: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

6 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

Israeli Labor Market, is a foundational paper for this study. Friedberg studied the Post-Soviet Aliyah as a natural experiment regarding the effect of the mass migration on native wage rates. Friedberg focused on industries employing high volumes of Soviet im-migrants relative to other industries. Friedberg used an Instrumental Variables approach; and uncovered a positive correlation between the presence of immi-grants in a profession in the Soviet Union and their presence in the same profession following migration to Israel. Friedberg shows that skilled Russian im-migrants were able to eventually find skill appropri-ate labor in Israel. Friedberg also determined that natives in occupations that came to employ high volumes of Soviet migrants experienced slower wage growth relative to natives in other industries. While this was considered to be evidence of the mi-gration’s effect on the labor market, the correlation was seen as relatively weak. However, Friedberg also found that in the period following the migra-tion, Israeli students were less likely to study medi-cine or engineering, the two industries most highly concentrated with Russians, at the university level. At the same time, natives became more likely to en-roll in legal education programs, a field with limited Russian presence. This shows that the migration impacted the labor decisions of native Israelis, and thus the labor market5. Mass Migration to Israel and Natives’ Employ-ment Transitions, by Sarit Cohen-Goldner and M. Daniele Paserman is another foundational paper for this study. Published in the Cornell University In-dustrial & Labor Relations Review, Cohen-Goldner & Paserman sought to determine how the Post-Sovi-et Aliyah affected the likelihood of native Israelis to transition from unemployment to employment. Due to the unique conditions of the migration (educat-ed immigrants, large volume, small state), Cohen-Goldner & Paserman believed that the Post-Soviet Aliyah may provide the “upper bound” for the po-tential effects of immigration shocks on native em-ployment transitions. Cohen-Goldner and Paserman determined that the immigration affected employ-ment to non-employment transition probability by a small margin of .49%. Additionally, no significant link was found between the immigration shock and the ability of natives to find work. The lack of a link between immigration and native’s ability to find work is consistent with studies performed in other states, and thus is not an indicator that the migration did not affect the Israeli labor market. According to Cohen-Goldner & Paserman, this could be because workers are most likely to transition to non employ-ment as a result of adverse business environments,

bankruptcy or employee dissatisfaction; none of which are related to labor markets. Technology, Trade & Adjustment to Immigra-tion in Israel by Neil Gandal, Gordon Hanson and Matthew Slaughter is was very influential on this work. Gandal, Hanson & Slaughter’s\\\\\\ paper sought to determine how the Israeli economy ab-sorbed the Soviet immigration shock. Gandal, Han-son & Slaughter’s work is very close to this study, as one of the possibilities explored is a scenario in which the immigration shock was absorbed by changes in output mix. Gandal, Hanson & Slaughter did this by examining the production aspect of the Heckscher-Ohlin model in order to determine which aspects of production absorbed the labor shock. According to Gandal, Hanson & Slaughter, output mix did shift towards high skill intensive produc-tion, but improved production techniques spawned by technological advances were more likely to have absorbed the labor market shock than changes in output mix. However, Gandal, Hanson & Slaughter do not establish causal relationships, and crucially leave open the possibility that Russian immigration produced endogenous effects that altered output mix. An example given is that Israel may have expe-rienced greater foreign capital inflows as a result of the migration, which would in turn foster progress in industries with high concentrations of Russian immigrants. Additionally, this study only uses data from 1989-1996. The period of growth for high-skill intensity sectors does not appear to stabilize until 2000, which leaves open the possibility that Gandal, Hanson & Slaughter may not have utilized the full scope of output mix changes6. The final influence to be discussed in this sec-tion is The Absorption of Highly Skilled Immi-grants: Israel, 1990-1995, by Zvi Eckstein and Yor-am Weiss. Eckstein & Weiss set out to determine wage trends and occupational transitions in Israel following the Post Soviet Aliyah. This paper con-firmed the work of Weiss, Sauer & Gotlibovski in 2000, which determined that immigrants take time to gradually adjust to new languages, market sys-tems and institutions; as well as to accumulate expe-rience and find appropriate employers. Thus, it takes immigrants time to reach full productivity. Eckstein & Weiss found that in immigrants to Israel in the 1990‘s experienced occupational downgrading, par-ticularly in the first three years. Following this pe-riod, the same immigrants began to find positions that matched their skill set and education level. This has important consequences for the effect of immi-gration on output mix. As a result of the transition period, changes in output mix can not be expected

Page 7: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 7

Volume xxiii. issue i.

to begin until 1996 at the earliest, which marks three years from the peak of Russian migration7.

IV. Theoretical Framework

1. Heckscher-Ohlin Model Streaming had boosted subscriber adds to lev-els tThe theoretical framework for this study lies in the Heckscher-Ohlin Model of production. Under Heckscher-Ohlin, relative endowments of factors of production dictate which industries in a country hold comparative advantage. The model depends on certain assumptions, such as a competitive mar-ket. In this case, the factors of production consid-ered will be high skilled and low skilled workers. As a result of Russian migration, large amounts of high skilled workers entered the country. High skill workers became more abundant than low skilled workers as a share of the total work force relative to pre immigration levels. Thus, we would expect to see Israel develop relative comparative advantage in high skilled sectors, as the factors of production for these sectors are now more abundant. Compara-tive advantage manifested itself as increases in the output and exports of high skill intensive sectors such as high skill manufactured goods, electrical equipment, scientific equipment and telecommuni-cations. Additionally, Heckscher-Ohlin states that changes in production offset wage variance due to factor changes. The factor change in this case is rep-resented by immigration.

2. Rybczynski Theorem Today, Netflix maintains competitive advantag-es agThe Rybczynski Theorem is a key consequence of the Heckscher-Ohlin Model. The suddenness of Russian migration to Israel makes this an ideal case to apply Rybczynski. The Theorem states that when the factor constraint for one factor of production shifts left, the growth in production of the now fac-tor abundant good will be greater than the fall in production of the good that did not receive an fac-tor inflow. In this study, Israel experienced a large shift in high-skilled labor. Thus, the Rybczynski Theorem implies that there will be an increase in the production and overall output of goods that require high skilled labor as a factor input. Additionally, there will be a smaller, but still substantial decline in the production of goods that are dependent on low skill labor as a factor input, as low skill labor is less abundant relative to high skilled labor following the period of Russian immigration. Figure 4 shows a model of the Rybczynski Theorem.

3. Heckscher-Ohlin Theorem The Heckscher-Ohlin Theorem is a conse-quence of the Heckscher-Ohlin Model of Produc-tion. This theorem considers the consequences of the model for profit maximizing firms. Following an inflow of factors, the goods created with said factors become cheaper to produce relative to other goods, and a country holds a relative comparative advan-tage in the production of these goods. In a market economy, profit-seeking firms will then begin to produce the now relatively cheaper good at a higher rate. Since the factor abundant goods are relatively inexpensive and are produced at a higher rate, the Heckscher-Ohlin Theorem states that the country will begin to export newly factor abundant goods at a higher rate; reflecting the newfound compara-tive advantage. The model implies that Israel should begin exporting high-skill intensive manufactured goods at a higher rate following the Post-Soviet Ali-yah.

V. Hypothesis

The Post-Soviet Aliyah presents an ideal op-portunity to test the effectiveness of the Heck-scher-Ohlin Model of Production and the ensuing Heckscher-Ohlin and Rybczynski Theorems due to the migrant’s high education quotient and the small size of the Israeli economy. This paper will exam-ine changes in output and export mix as a means of measuring the effect of the migration shock. The hypothesis, based on the Rybczynski Theorem is as follows; following Russian immigration in the early 1990’s, Israel will see an increase in the production of high skill intensive manufactured goods, and a less than proportionate decrease in low skill inten-sive goods. Additionally, Israel will begin to export more high skill intensive goods as a share of total exports. This hypothesis is based on the migrants’ high education quotient, as well as their likelihood to be employed in skill intensive sectors. The immigra-tion shock should induce growth in the Israeli high-skill sector. This study differs from Gandal, Hanson & Slaughter, as it will consider output and export data until 2000. As the bulk of the migration oc-curred in the early 1990’s, considering output data until 2000 allows for a lag period between the mi-gration and observable effects. The lag period ac-counts for various adjustments, such as the transi-tion to the Hebrew language and the psychological and cultural challenges associated with transitioning from a communist state to a capitalist, free-market economy.

Page 8: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

8 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

VI. Research Method

Data regarding immigration levels was found in Tolts, and data regarding education levels was taken from Cohen-Goldner/Paserman, and original-ly found in the Israeli Labor Force Survey. Beyond this, the majority of the data was taken from from the OECD’s Structural Analysis, or “STAN” data-base or the World Bank. The STAN database was used to mine both export and output data, while the World Bank was used for Gini coefficients. Due to data limitations, this study focused on the manufac-turing sector’s output mix. Gross output is defined as value in current US Dollars. This study ana-lyzed data from 1990 to 2000. 1990 was used as the benchmark for pre-Russian immigration Israel. An upward trend in high skill production was expected between 1996 and 1998, which allows for a 3-5 year adaptation period for immigrants following the bulk of the migration. This study used OECD aggregated statistics for both high and low skilled intensive produc-tion. The STAN measurements used were aggre-gated Low Technology Manufactures (LOTECH) and High & Medium-High Technology Manufac-tures (HMHTECH). Both statistics are aggregated using ISIC Revision 3’s definition of technology intensity, which categorizes industries by R&D in-tensity. LOTECH includes low skill manufactures, paper products, food products, tobacco and textiles. HMHTECH aggregates various high skill manu-factures, such as electrical machinery, telecommu-nications equipment and pharmaceuticals. Then, both LOTECH and HMHTECH were viewed as a percentage of overall manufacturing output to deter-mine which aggregation grew relative to the other. This analysis was used to prove the relevance of the Rybczynski Theorem, and the immigration shock’s effect on output mix. High and Medium High classi-fications were considered together, as both included industries that saw input factor shocks during the period of Russian migration. As long as the aggrega-tion of both holds a higher percentage share of Israe-li manufacturing output after the immigration shock than before, the Rybczynski Theorem will have held true8. To prove the Heckscher-Ohlin Theorem, an identical analysis was performed using LOTECH and HMHTECH exports as a share of total exports; rather than as a share of overall production. It is notable that services were excluded from the study. Traditionally, services make up a signifi-cant portion of output in both the high and low skill intensive sectors. However, sufficient data for Israeli service production or exports was not available for

the time period required, and as a result, only man-ufacturing was considered. The study would have expected high skill services, like high skill manu-facturing production, to increase at a greater than proportional rate to a subsequent fall in production of low skill intensive services as a share of over-all service production. However, it is also expected that the magnitude of the change would be slightly less than the change in manufacturing to account for adaptation to the Hebrew language, as services is traditionally more language intensive than manufac-turing.

VII. Changes in Output Mix

An analysis of manufacturing output data sup-ports the hypothesis. In 1990, prior to the immigra-tion shock, low skill manufacturing output made up 49.18% of overall manufacturing output; and high and medium high skill manufacturing output made up 32.88% of overall output. In 1993, the peak of migration, these figures are essentially constant. At this time, low skill production made up 50% of overall manufacturing output, and high skill manu-facturing made up 32.84% of output. In 1996, after the end of the lag period, the high skill share begins to trend upwards, while the low skill output share begins to decrease. This can be observed in the graph at Figure 5. In 1996, high skill output stood at 34.25% of overall output, and low skill output stood at 47.63% of overall output. By 2000, most migrants adapted to Israeli society and began to affect eco-nomic output, high skill manufacturing output made up 44.07% of overall manufacturing output, while low skill production made up just 40.74% of overall output. The change in output mix during the latter part of the 1990’s displays a large increase in high skill manufacturing output, and a less than proportional decrease in low skill output. This is consistent with the implications drawn from the Rybczynski Theo-rem. During the 1990’s, high skill output increased by 34.03%, and 11.19% of total manufacturing pro-duction. During the same period, low skill output decreased by 17.16%, and 8.44% of total manufac-turing production. This is consistent with the Rybc-zynski Theorem, and indicates that changes in factor inputs were likely the cause of output mix changes. The fastest growing sectors during this period were electrical equipment and optical equipment. By the year 2000, electrical equipment production had grown by 274.31% from the peak of the migra-tion. Electrical and optical equipment production also grew by 127.97% from 1996, the year in which

Page 9: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 9

Volume xxiii. issue i.

the majority of migrants began to find skill appropri-ate labor. Electrical equipment production’s share of overall manufacturing rose to 23.97% in 2000, from just 15.56% in 1996. This represents an increase equivalent to 8.41% of total manufacturing output. The slowest growing sectors were food products, beverages and tobacco, designated by the OECD as a low skill input industries. As this was the slowest growing sector, food production’s share of overall production declined relative to other sectors. Prior to the immigration shock, food was Israel’s largest low skill input sector by overall production. Food production’s share of overall GDP fell by about 1.65%, which represents a decline that is less than proportional to the increase in electrical and optical equipment. The Heckscher-Ohlin Theorem states that in a free economy featuring profit seeking firms, coun-tries experiencing a factor supply shock will see ex-ports of factor abundant goods increase at a greater than proportional rate to exports of goods that are now relatively less factor abundant. This trend is shown in Figure 6. As Israel is a capitalist economy featuring profit seeking firms, firms began to export more high skill intensive or technology goods as their comparative advantage grew. Between 1996 and 2000, high skill intensive exports increased by 16.9%, and by 8.4% of total manufacturing exports. Simultaneously, low skill intensive and labor inten-sive exports decreased by 17.5%, and by 7.5% of overall manufacturing exports. This trend is consis-tent with the Heckscher-Ohlin Theorem, and indi-cates that Israel was gaining a relative comparative advantage in high skill intensive manufacturing goods during the years following Russian migra-tion. It is notable that low skill goods fell at a great-er rate than was expected. This is likely due to the expiration of the Multi-Fibre Arrangement in 1995, which will be discussed with further detail in a later section. If the immigration shock affected output, there must be evidence that immigrants found work in high skill sectors. Eckstein & Weiss proved that Russian immigrants eventually found skill appro-priate labor by observing large skill upgrading a few years following the migration. The presence of skill upgrading indicates that immigrants began to work in high skill sectors and in positions that matched their education level. The presence of immigrants in high skill sectors indicates that migrants were work-ing in a capacity from which they can affect produc-tion patterns6. According to Eckstein & Weiss, just 18.7% of arriving Russian engineers were employed in a “high-skill” occupation after 6 months, but af-

ter four years this figure rose to 42.6%. The trend of skill upgrading explains why production patterns began to change in 1996, a few years after the brunt of the migration. Additionally, Gandal, Hanson & Slaughter’s paper supports a similar conclusion. Gandal, Hanson & Slaughter found that unemploy-ment rates for Russian immigrants tripled between 1989 and 1991. However, by 1996, the year the im-migrants are assumed to have reached a state of rela-tive economic assimilation, unemployment rates for Russian immigrants were within just two points of native Israelis7. This shows that the immigrants be-gan finding work, and thus affecting the economy and production patterns. The successful assimilation of Russians into the work force can be attributed to strong public institu-tions such as the Ministry of Immigration Absorp-tion (MIA), and The Ministry of Education. The MIA encourages economic assimilation by organiz-ing Hebrew classes and offering stipends for lan-guage training; as well as by matching qualified im-migrants with capital for start up businesses, while the Ministry of Education maintains a Department for the Evaluation of Foreign Degrees. Additionally, the Center of Absorption of Science works to match immigrants with scientific backgrounds with Israeli employers7. The effort of the Israeli state to match skilled immigrants with high skill work placed edu-cated and skilled Russian immigrants in a positions to succeed and influence the economy, and thus con-tributed to the growth of Israeli high skill produc-tion during the late 1990’s.

VIII. Stratification

The effects of rapid modernization are not ex-clusively positive. While the Post-Aliyah brought many positive changes to Israel, the rapid growth of the economy brought increased stratification.

IX. Changes in Productivity Per Worker

In the late 1990’s, worldwide demand for technol-ogy exploded. In particular, demand for telecom-munications equipment increased rapidly as a result of the new cell phone and internet industries. Fol-lowing the demand shock, many developed nations began to produce more high skill intensive goods.. One potential confounding explanation for the trends found in this study is that the technology de-mand shock drove the changes in Israel’s production patterns. However, Figure 7 refutes this. Figure 7 shows output per high skilled worker in Israel in the 1990’s. Productivity per worker rises following the

Page 10: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

10 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

immigration shock. Rising productivity per worker shows that firms were becoming more effective and efficient at producing technology, most likely as a result of the influx of high skilled labor. The in-creased efficiency and productivity rates of Israeli workers indicates a growing comparative advan-tage. A growing comparative advantage indicates that Israel’s production shifts and increased high tech output were not driven exclusively by global increases in demand for technology.

X. Stratification

The effects of rapid modernization are not ex-clusively positive. While the Post-Aliyah brought economic growth, this period also brought increased stratification. Increased stratification was observed through an analysis of changes in Israel’s Gini coef-ficient during the 1990’s. Gini coefficients are used to measure economic equality, with a higher number indicating greater inequality. Over the 1990’s, as the Post-Soviet Aliyah began to affect society, Israel’s Gini coefficient rose. Limited statistics exist on Gini coeffecients during this period, but figures are avail-able for 1993, the year the migration began to affect the Israeli economy, and 2001. Between 1993 and 2001, Israel’s Gini grew by almost 3%, a substantial margin. This indicates that Israeli society became more unequal as the effects of the Post-Soviet Ali-yah took hold9.

XI. Evolving High Skill Employment Levels The expiration of the Multi-Fibre Arrangement, or MFA, is another potential confounding expla-nation for a shift in production patterns. The MFA imposed quotas on the amount developing countries could import to developed countries. This was in-tended to protect industries in developed nations, which were very vulnerable to developing nation’s absolute advantages in low skill intensive manu-facturing. When this expired, developing countries began to export large amounts of low skill intensive goods as a result of their absolute advantage in this area due to minimal labor costs. It is likely that this helped reduce Israel’s overall exports of low-skill intensive goods, such as textiles and food, due to Israel’s decline in competitiveness. However, as is shown in Figure 8, the overall number of high skill employees in Israel rose over the 1990’s. This ef-fect is consistent with the Rybczynski Theorem, and shows that the Israeli high technology sector is drawing in employees from other sectors. This trend also indicates that Israel’s high skill sector was

evolving and displaying a growth trend independent of any changes regarding low skill intensive manu-facturing. While it is very possible and even likely that the expiration of the MFA affected Israeli ex-ports, the upward trend in high skilled employment indicates that changes in Israeli production patterns were not exclusively a result of the expiration of the MFA, and the period of Russian migration was likely a major factor.

XII: Conclusion

It is clear that the inflow of high skilled labor created by Russian migration to Israel affected the output mix of the Israeli economy. The trend of growth in high skilled output and exports aligned with a less than proportional decreases in the output and exports of low skill intensive goods relative to overall production confirms the Rybczynski Theo-rem. It is crucial that changes in output mix and ex-port mix began 1996, 3 years after the majority of the migration occurred. The presence of a lag period aligns with Eckstein & Weiss, who found that Rus-sian immigrants to Israel in the early 1990’s took about three years to find skill appropriate labor due to the language barrier and new labor market insti-tutions. Further supporting the original hypothesis, changes in high skill employment and production per worker indicate that the high skill manufactur-ing sector in Israel was affected by the migration. The mechanism with which immigration af-fected output mix is unclear, as many possibilities exist. One of these is that the immigration shock made R&D intensive firms more efficient. Another is that the high-skill labor inflow endogenously en-couraged technologic growth, which made native employees of static skill levels more productive. Finally, it is also possible that immigration caused capital inflows from the migrants’ origin country, which increased firm productivity. Regardless, it is clear that the immigration shock induced growth in high skill intensive production and exports, and it is likely that the same effect would be seen in other economies.

XII. Policy Implications

Production patterns following Russian migra-tion in the early 1990‘s display a positive correla-tion between inflows of high skilled labor and the production of high skill intensive goods. As a result of the small size of the country and the massive immigration shock, Israel likely represents the up-per bound of this effect. However, the presence of

Page 11: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 11

Volume xxiii. issue i.

51-65. Print.

“Structural Analysis Database.” . OECD, 1 Jan. 2014. Web. 3 Mar. 2014. <http://www.oecd.org/sti/ind/48350231.pdf>.

Tolts, Mark. “ Post-Soviet Aliyah and Jewish De-mographic Transformation.” . Academia.edu, 2 Aug. 2009. Web. 14 Feb. 2014.

“World Bank Gini Index.” The World Bank IBRD. The World Bank, 1 Jan. 2014. Web. 4 Nov. 2014. <http://data.worldbank.org/indicator/SI.POV.GINI?page=3>.

XV. Footnotes1. Israel Central Bureau of Statistics - Found Through Jewish Virtual Library, 1 Jan. 2014. Web. 10 Apr. 2014.

2. Shuval, Judith, Sergio DellaPergola, and Elazar Lesem. Immigration to Israel: Sociological Perspec-tives. Livingston, NJ: Transaction Publishers, 1998. 51-65. Print.

3. Tolts, Mark. “ Post-Soviet Aliyah and Jewish Demographic Transformation.” . Academia.edu, 2 Aug. 2009. Web. 14 Feb. 2014.

4. Cohen-Goldner, Sarit, and M. Daniele Paserman. “Mass Migration to Israel and Natives’ Employment Transitions.” Industrial and Labor Relations Review 59: 630-652. Web. 13 Feb. 2014.

5. Friedberg, Rachel. “The Impact of Mass Migra-tion on the Israeli Labor Market.” The Quarterly Journal of Economics 116: 1373-1408. Web. 17 Feb. 2014.

6. Gandal, Neil, Gordon Hanson, and Matthew Slaughter. “Technology, trade, and adjustment to immigration in Israel.” European Economic Review 48: 403-428. Web. 17 Feb. 2014.

7. Eckstein, Zvi, and Yoram Weiss. “The Absorption of Highly Skilled Immigrants: Israel, 1990-1995.” Tel Aviv University & Boston University 48: n. pag. Web. 17 Feb. 2014.

8. “Structural Analysis Database.” . OECD, 1 Jan. 2014. Web. 3 Mar. 2014. <http://www.oecd.org/sti/ind/48350231.pdf>.

the trend indicates that countries desirous of larger high-skill sectors, such as technology or financial services should seek out high skilled labor from oth-er countries. Attracting high skill labor is a possible solution for developing countries that would like to alter production patterns and diversify outputs and exports. A mechanism for attracting high skilled workers could take numerous forms. One potential solution is program based around research grants designed to entice academic professionals. Another is offer-ing special tax considerations to draw high skilled private sector labor from other countries. One way for countries to potentially import high skill labor in large quantities would be to offer incentives for R&D intensive firms. Tax breaks for R&D or other highly skilled branches of firms would draw large amounts of high skilled labor at once, which would help developing countries increase output and ex-ports of high skill intensive goods. It is also key to note that Gandal, Hanson & Slaughter left open the possibility that high skill labor inflows endogenous-ly encourage technology change, which then has the potential to beget further growth. This would be a key area for future research to focus on.

XIV. References

Cohen-Goldner, Sarit, and M. Daniele Paserman. “Mass Migration to Israel and Natives’ Employment Transitions.” Industrial and Labor Relations Review 59: 630-652. Web. 13 Feb. 2014.

Eckstein, Zvi, and Yoram Weiss. “The Absorption of Highly Skilled Immigrants: Israel, 1990-1995.” Tel Aviv University & Boston University 48: n. pag. Web. 17 Feb. 2014.

Friedberg, Rachel. “The Impact of Mass Migration on the Israeli Labor Market.” The Quarterly Journal of Economics 116: 1373-1408. Web. 17 Feb. 2014.

Gandal, Neil, Gordon Hanson, and Matthew Slaugh-ter. “Technology, trade, and adjustment to immigra-tion in Israel.” European Economic Review 48: 403-428. Web. 17 Feb. 2014.

Israel Central Bureau of Statistics - Found Through Jewish Virtual Library, 1 Jan. 2014. Web. 10 Apr. 2014.

Shuval, Judith, Sergio DellaPergola, and Elazar Le-sem. Immigration to Israel: Sociological Perspec-tives. Livingston, NJ: Transaction Publishers, 1998.

Page 12: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

12 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

12. Eckstein, Zvi, and Yoram Weiss. “The Absorp-tion of Highly Skilled Immigrants: Israel, 1990-1995.” Tel Aviv University & Boston University 48: n. pag. Web. 17 Feb. 2014.

9. “World Bank Gini Index.” The World Bank IBRD. The World Bank, 1 Jan. 2014. Web. 4 Nov. 2014. <http://data.worldbank.org/indicator/SI.POV.GINI?page=3>.

XVI. Tables & Figures

Figure 1:

Figure 2:

Figure 3:

Figure 4:

Figure 5:

Figure 6:

Figure 7:

Page 13: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 13

Volume xxiii. issue i.

Figure 8:

Page 14: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

14 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

I. Introduction

Introduced in 1989 at the rate of 3%, the con-sumption tax in Japan is implemented in the form of Value-Added Tax (VAT). As consumption tax, VAT differs from sales tax in that it applies to the market value added of a product at each stage of manufac-ture or distribution as opposed to final sale. The conventional sales tax is applied only to the final consumer of the product or service, which creates incentives for tax-evading activities by pre-tending to be businesses in the middle of the supply chain. The VAT eliminates the tax-evading opportu-nity since it applies to the “value-added” (difference between sales price and input price) at every stage along the supply chain. Moreover, the mechanism of VAT practically encourages businesses at later stages of the supply chain to prove that the upstream business (the business they buy inputs from) has paid its VAT so that they can pay their own VAT at a lower rate. In this way, businesses are essentially monitoring each other to comply. Despite some ad-ditional accounting costs, above features all make the VAT a more efficient means of generating tax revenue. Though the VAT was hardly known to pub-lic in the early 1960s, it is now implemented in over 150 countries around the globe, generating 20% of the world’s tax revenue and influencing around 4 billion people. Its effectiveness in raising tax rev-enue and relatively low administration costs had made VAT a popular means of taxation for countries with all levels of living standards (Keen & Lock-wood, 2010). Compared to other major economies, Japan in-troduced the VAT rather late (not until 1989). While Japan’s public debt has been rising steadily over

the past two decades (currently over 160% of GDP) primarily due to the increase in social security pay-ments and shrinking in the labor force, Japan’s VAT rate of 5% is the lowest among all OECD countries . Raising the VAT rate has been long-viewed as a via-ble and efficient way to restore Japan’s fiscal health, since both theoretical and empirical evidence sug-gest that the VAT provides a stable source of tax rev-enue in an aging society like Japan (given consump-tion is smother than income across generations). Moreover, the tax burden would be divided more equitably across current cohorts after the VAT hike because those who entering retirement would pay a fairer share toward the cost of their retirement sup-port. Despite above benefits for raising consump-tion tax, critics of VAT argue that the flat-rate tax is a de facto regressive tax since it places relatively higher burden on the low-income households who presumably spend higher percentage of their income on consumption. However, significant evidence has shown that the VAT is substantially less regressive when viewed from a lifetime perspective and adjust-ments such as zero-rating on essential products in-cluding food and medicine would reduce its regres-sivity problem even further. (Caspersen & Metcalf, 1995) In August 2012, after years of political negotia-tions, both Houses of the Japanese Diet passed the bill to raise the nation’s VAT rate to 8% by April 2014 and to 10% by October 2015. The decision to raise VAT rate met considerable opposition in Ja-pan, as some feared that the consumption tax hike would distort domestic consumption and put an end to Japan’s recovering trend in 2013, even drag the nation to a recession just like the one in 1997 when VAT rate was raised from 3% to the current rate of

Sui ZhangPurdue University

The Impacts of VAT Hike on the Japanese Economy

EditorialSui Zhang of Purdue has written an excellent analysis on the Value Added Tax (VAT) in Japan. The Japanese economy presents an interesting case study for a tax on consumption. Long struggling with deflationary conditions and slowed by a remarkably high saving rate, Japan has a much lower consumption rate than many western economies like the United States. Manipulation of the Value Added Tax in Japan allows for an interesting opportunity to observe the effects throughout the economy.

Jacob Miller

Page 15: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 15

Volume xxiii. issue i.

5%. However, the fact that the consumption tax hike in 1997 coincided with the Asian Financial Crisis makes it difficult to isolate the real effects bought by the tax hike. My paper aims to evaluate the macroeconomic impacts of Japan’s past consumption tax hikes on its economy both theoretically through simulation of model and empirically through regression analysis in comparison with South Korea . Though Japan and South Korea experienced distinct growth patterns in the past few decades with the Korean economy expanded rapidly while Japan suffered from weak growth, both countries are now similar in standard of living and experienced identical major inter-national shocks. Moreover, the VAT rate in South Korea stayed constant at the level of 10% since its implantation in 1976 which makes the country a suitable comparison to contrast the impact of VAT hike on the economy . Internationally, adjusting VAT rate has gained much popularity among policy-makers: in terms of both VAT hike to improve fiscal condition (as the case in Japan) and VAT cut as stimulus measure (as the case in the United Kingdom). Regardless of the level of VAT rate, a uniform rate system with a broad tax base, few exemptions and modernized procedures for enforcement and compliance is widely favored in economic literature. Nishiyama (2012), Rao (2013), Bye (2012), Claus (2013) and Rawat et al. (2010) discussed the welfare superiority of a single-rate VAT system to a varying-rate system in detail. Bhattarai (2006) used a micro-founded General Equilibrium Model with labor market only to study the impacts of consumption tax versus that of labor income tax. The simulation results showed that un-der revenue equivalence hypothesis, the distortion effect of consumption tax is less significant than that of labor income tax. Thus replacing labor income tax with revenue-neutral consumption tax would in-crease the social welfare. Fernandez (2011) applied a dynamic Gerneral Equilibrium Model to study the temporary VAT cut policy implemented in the United Kingdom. Since the duration of a tax cut is pre-determined and announced publicly, the impacts on the economy when the VAT rate returns to nor-mal would be equivolent to a VAT hike. The empri-cal evidence in U.K. suggested that the consumption and output level after the tax cut expired was lower than that during the tax cut and the stedy state level. Studies on the U.K. VAT cut by Blundell (2009) and Crossley et al. (2009) also replicated similar find-ings. Miki (2011) studied the impacts of VAT rate change on aggregate consumption and economic

growth by performing multiple regressions on em-pirical data from 14 developed countries from 1980 to 2010. Results revealed that both the consumption and growth level tend to be higher immediately be-fore the VAT hike and drop dramatically after the hike. A gradual recovery trend was also observed when a sufficient amount of time had passed since the tax hike. The following section presents the detailed the-oretical model derived to study the impacts of VAT hike on macro economy. Section III describes the source and characteristics of empirical data. Section IV presents the empirical strategies used to test the model developed in section II. Section V discusses results and findings from both simulation and em-pirical estimation and section VI concludes the pa-per.

II. Model

This paper considers a micro-founded Gen-eral Equilibrium Model originally motivated by Blanchard and Fischer (1989) and modified by Bhattarai (2006) with inclusion of the effects of consumption and labor income taxes. In addition to the original model, I added the role of capital market and government spending in determining the gen-eral equilibrium. This modification is realistic and relevant in that the change in consumption tax rate would inevitably alter the household’s consump-tion-saving behavior which could have subsequent effects on production through the capital market. Further justifications for this modification are given by Bobenrieth et al. (2008) The macro economy is simplified into three sectors: households, firms and the government. Representative households maximize utility by con-suming goods and services provided by firms and enjoying leisure time. In addition to a fixed income endowment, the households receive income from the return of labor in the form of wage payments and the return of capital in forms of firm profits and in-terest payments from savings. Representative firms use labor and capital to produce goods and services and aim to maximize profit which is the difference between revenue and cost of inputs. The govern-ment collects taxes to make transfer payments and provide public goods and services. Some underly-ing assumptions of the model include: (i).House-hold savings becomes investment as capital input to production and the returns on investment are costs to firms; (ii).The government eventually returns all tax revenue to households as lump sum transfer pay-ments; (iii).All fiscal and monetary policy changes

Page 16: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

16 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

are considered exogenous shocks.

As illustrated in Figure.1, the collective optimiza-tion decisions made by individual households and firms lead to a general macroeconomic equilibrium in which perfect match is achieved between demand and supply for final goods and services as well as production inputs (labor and capital). At the general equilibrium, the set of prices and quantities elimi-nates any excess supply or excess demand that all outputs supplied by firms are consumed by house-holds, such equilibrium is stable in the long run given no outside shocks.

A. HouseholdsThe representative household aims to maximize the utility which is determined by a Cobb-Douglas func-tion of the amount of consumption spent on goods and services and the amount of leisure time enjoyed:

where c is the amount of consumption spent on goods and services and l is the amount of leisure time; ϕ is the weight of consumption and (1-ϕ) is the weight of leisure time in utility determination.The representative household is subject to the three following constraints:where π is profit from owning a firm; is the amount

of income endowment ; ks is the amount of money put into savings; kd is the amount of capital demand-ed by the firm; p is the price level of final goods and services measured in terms of capital goods and w is the wage rate measured in terms of capital goods. r is the interest return on investment; hs is amount of labor hour supplied. The time endowment is nor-malized to 1 for simplicity. Equation (3) represents the budget constraint that a representative household faces without the

presence of consumption and income taxes. Now we consider the case where taxes are present. Ac-cording to arguments by Alm and Ganainy (2012), VAT is essentially equivalent to retail sales tax since it taxes on the value-added of all businesses from all production stages which equals to the total value of all final products. As other forms of taxes on firms, the burden of VAT will be transferred along the sup-ply chain eventually to the final consumer in terms of higher sale prices. Therefore, after the implemen-tation of consumption and income taxes, the house-hold budget constraint (3) changes to:

where tc is consumption tax rate and th is the labor income tax rate; R is tax revenue.The Lagrangian function for household utility opti-mization then changes to:

By solving the constrained maximization of (6) sub-ject to (2),(4),(5) by setting First Order Conditions equal to zero, we have the after-tax optimal labor supply and consumption level as:

In order to find the optimal level of demand for labor and capital and profit, we need to consider the optimization problem in the firms sector.

B. Frims The representative firm aims to maximize the profit which is the difference between revenue gen-erated by supplying goods and services given cur-rent technology and levels of inputs and the costs of inputs:

where yd is the total output demanded by the market (consumed by households); hd is the amount of labor hour demanded; r is the rental rate of capital and kd is the same as above which denotes the amount of capital demanded by the firm. The representative firm is subject to the two fol-

Page 17: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 17

Volume xxiii. issue i.

lowing constraints:

where ys is the total output supplied by the firm; A is the total factor productivity and is normalized to 1 for simplicity; α is the output elasticity of labor and (1-α) is the output elasticity of capital; Constant return to scale (RTS) is assumed.At the general equilibrium, the equilibrium price level eliminates excess demand or excess supply in the output market, and since household consump-tion constitutes all the aggregate demand, the fol-lowing conditions are satisfied:

The Lagrangian function for profit maximization by firm is given by:

By solving the constrained optimization of (13) subject to (10) and (11) using similar techniques as above and applying the output equilibrium condi-tion of (12), we have the optimal levels of labor de-mand as:

The general equilibrium level of profit can then be written as:

To solve the equilibrium level of above variables explicitly, we introduce the final sector of the econ-omy: the government.

C. Government

The government collects consumption tax from the output market and labor income tax from the labor market. The government budget constraint is given by:

Where c$ and h$ are market-clearing levels of out-put/consumption and labor hours respectively. By substituting (16) into the (7) and (8) and apply-ing the equilibrium conditions for labor and capital market, we can solve for the ratio between equilib-rium wage rate and rental rate of capital as:

Hence, the equilibrium level of output and con-sumption can be solved explicitly:

And the equilibrium level of labor supply is given by:

All other endogenous variables can then be ex-pressed explicitly in terms of parameters using the relationships above.

III. Data

In order to test the above model empirically, the quarterly data for major economic indicators are collected for Japan and South Korea from the period 1980 Quarter 3 to 2013 Quarter 4 (134 pe-riods). Macroeconomic indicators including Gross Domestic Product (GDP), Consumer Price Index (CPI), Producer Price Index (PPI), Hourly Earnings and Monthly Hours for Manufacturing, total retail sales are collected from International Monetary Fund (IMF)’s International Financial Statistics and Organization for Economic Co-operation and De-velopment (OECD)’s Revenue Statistics. All above data are seasonally-adjusted to remove any persis-tent seasonal effects. Real U.S. import price for petroleum data from the U.S. Energy Information Administration (EIA) are used as proxy variable for world energy price whose fluctuation is a major international shock for major net energy importers as energy input could alter certain technological constraints. In addition, the quarterly GDP growth rate for all 34 OECD countries are collected as proxy for other interna-tional shocks and development trends. Binary vari-ables for the 1997 Asian Financial Crisis and the Financial Crisis of 07/08 are also included. Descriptive statistics for key domestic variables are shown in Table.1. Real private sector earnings is calculated by dividing real private sector earnings

Page 18: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

18 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

by the corresponding price level (CPI).

Summary statistics for key international shock vari-ables are shown in Table.2

As shown in Figure 2, the time frame of the sample data covers both of Japan’s past consump-tion hikes (1989 Q2 and 1997 Q2) while Korea’s VAT rate remained 10% throughout the frame.

Thus the data sample from Korea serves as a control group to model an economy without consumption tax shocks and isolates the real effect of consump-tion tax hikes on Japanese economy. Compared to the data used by Miki (2011), this sample covers more recent observations, includes country without VAT rate change as control group and uses addition-al proxy variables for international shocks.

IV. Empirical Strategy

A. Simulation of Model Before examining the macroeconomic impacts of consumption tax hikes on Japanese economy us-ing the empirical data collected above, I construct two types of simulation of the model developed in Section II. The first type of simulation is the base scenario where a reasonable set of preference and technol-ogy parameters given in Table.3 is chosen to com-pare the equilibrium levels of utility, total output, labor supply and real wage rate given different levels of consumption tax rates. I will first exam-ine the simulation results of the five static scenarios where consumption tax is held at 0%, 3%, 5%, 8% and 10% respectively, which mimic the steps of Ja-pan’s past and future consumption tax hikes. Rela-tive change in levels against those in the scenario where no consumption tax is implemented will be studied to reflect the magnitude of impacts bought by implantation of consumption tax. Stepwise rela-tive changes will also be considered to capture the marginal effects of each step in the consumption tax hike. Supplemental to the stepwise approach, I will then allow consumption tax rate to move in small increments from 0% to 10% to test the sensitivity of above macroeconomic variables to consumption tax rates. This would be similar to an impulse response analysis except that the horizontal axis takes on val-ues of the consumption tax rate.

The second type of simulation includes a sensi-tivity analysis of the impact of consumption tax hike on endogenous variables to household preference and production technology. The analysis uses util-ity as the measure for social welfare and focuses on its sensitivity to the preference and technology pa-rameters. The model parameters are manipulated to move in increments and ranges specified in Table.4 and the magnitude of the consumption tax hike is set to be from 3% to 5%, mimicking Japan’s most recent tax hike in April 1997.

Page 19: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 19

Volume xxiii. issue i.

B. Multiple Regression To test whether the simulation results of model are consistent with empirical evidence from Japan’s two past consumption tax hikes, multiple linear re-gression will be performed on production of total industry and total retail trades. The models specified below will be estimated using the ordinary least square (OLS) technique for the time series data from Japan and Korea separate-ly. The level of VAT rate as well as the binary vari-able indicating the one-time tax hike are included as explanatory variables along with other control vari-ables for monetary policy and international shocks. The natural logarithm transformation will be ap-plied to the total output data as shown below:

where yt is the production in total industry at period t; β0 is the constant term; vathk is the binary variable taking on the value “1” for the periods of VAT hike and “0” otherwise; t is the time trend variable; vi,t is the group of all other control variables and βi is the slope; ut is the stochastic error term assumed to be independent of all explanatory variables. To better contrast the periods before and after the consumption tax hike, additional binary vari-ables indicating the periods surrounding the tax hike are applied. For instance, aj takes on the value “1” for periods j year(s) after the tax hike and bk takes on the value “1” for periods k year(s) before the tax hike, “0” otherwise. In my analysis, the three-year neighborhood around the tax hike is considered. Though Korea did not go through any consump-tion tax hike during the periods of interest, the same values of aj and bk as Japan’s will be attributed to Korea. This treatment essentially makes Korea the control economy without the tax hike and the differ-ence in slopes of these variables between Japan and Korea is notably effective in isolate the impacts of consumption tax hike on Japanese economy. More-over, the comparison with Korea also enable us to simulate what would have happened were there not a tax hike in Japan. Hence, the model below will be estimated supplemental to model (44).

where δj is the partial effect of j year(s) after the tax hike and ηk is the partial effect of k year(s) before the tax hike; other variables have the same interpreta-tion as above. The difference in these partial effects between Japan and Korea will be tested statistically through estimating model (46) using the combined data set

of both countries and interaction terms with country indicator variable. The t-test against the null hypoth-esis that each coefficient of interaction term is zero will be performed to conclude whether a significant difference exists between the two countries during periods surrounding the tax hike. Below is the mod-el to be estimated for the combined data set consist-ing both countries:

where jpn is the country indicator variable taking on value “1” for observations from Japan; and are slopes of the interaction terms whose signifi-cance will be tested. Models (44) to (46) and above techniques will also be applied when total retail trades is used as response variable. Lastly, since above regressions involve time series data, the Durbin-Watson test for AR(1) serial correlation and the Breusch-Pagan test for heteroskedasticity will be performed. Any addi-tional remedial measures will be made if necessary to achieve unbiased inference.

V. Findings

A. Simulation Results The model simulation results of the base sce-nario using the set of parameter values specified in Table.3 are summarized below in Table.5 and Table.6. Percentage change of levels against the scenario without consumption tax is obtained in Ta-ble.5 and the stepwise comparison results are shown in Table.6.

Simulation results shown in Table.5 reveal that the implementation of consumption tax has negative impacts on variables including utility, total output (consumption), labor supply, real profit and posi-tive impacts on variables including leisure, tax rev-enue and real wage rate. The magnitude of impacts on all variables is greater for higher consumption tax rates. The most dramatic distortion effect is ob-

Page 20: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

20 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

served in labor supply that the implementation of consumption tax at 10% lowers the amount of labor supply by 6.08%. The effects on both consumption (total output) and utility are negative as expected, but the effect on utility is not as significant (a 10% consumption tax only lowers utility by 0.58%). The power of consumption tax in generating tax revenue can be backed by the above simulation that the con-sumption tax at the level of just 3% would boost the tax revenue by more than 20%.

Results shown in Table.6 report the percentage change of variables compared to those at the previ-ous level of consumption tax rate. For example, the utility level at 5% consumption tax rate is compared to that at 3% rate and the same comparison has been drown for all other variables and tax rates. The step-wise comparison show similar findings as above and reveal the marginal effect of each step in the con-sumption tax hike. The marginal effect on variables when consumption tax rate is raised from 3% to 5% is comparable with that when the tax rate is raised from 8% to 10%. Similarly, the marginal effect of a 3% tax hike is comparable with that when an initial 3% tax rate is introduced. Such pattern indicates that the magnitude of impacts is proportional to the mag-nitude of tax rate change. The effect on variables such as consumption (total output), and tax revenue also exhibits a typical pattern of diminishing mar-ginal effect as exposed visually in Firgue.3.

In the base scenario where all parameters other than consumption tax rate are held constant, when VAT rate is raised from 0% to 10%, the relationship of utility, leisure (labor supply) and profit with VAT rate is asymptotically linear after standardization. This indicates that the marginal effect of tax hike on above variables is nearly constant. Meanwhile the relationship of consumption (total output) and tax revenue with VAT rate reveals a certain degree of di-minishing marginal effect of consumption tax hike. The other part of model simulation involves the sensitivity test of model results to varying pa-rameters. The household preference and production technology parameters as well as the initial income endowment are now manipulated to change in range and increments specified in Table.4. When one pa-rameter is manipulated to vary in value, all other parameters are held at their base scenario levels. Percentage changes of utility when consumption tax rate is raised from 3% to 5% at each level of param-eters are obtained in Table.7 to test the robustness of simulation results.

The results of sensitivity analysis shown in Table.7 indicate that both the direction and magnitude of impacts on endogenous variables are robust to the change in parameters (both below and above the base scenario level). It is noteworthy that the wel-fare costs of consumption tax hike is particularly inelastic to the level of income endowment (magni-tudes of impact at each endowment level are identi-

Page 21: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 21

Volume xxiii. issue i.

cal up to 4th decimal place), which implies that the marginal effect of consumption tax hike is similar in magnitude for households with different levels of income endowments.

B. Regression Evidence The OLS estimation results of model (44) and (45) with the natural logarithm of production of to-tal industry as response variable are summarized below in Table.8. Several key observations can be made from this set of regression.

As shown in Table.8, the effect of a one-time VAT hike on production level seems to be negligi-ble after other variables have been controlled for. However, the coefficients of the binary variables indicating periods surrounding the VAT hike reveal a different picture. The coefficients for periods be-fore the tax hike are all significantly positive for both countries, implying that the production level during those periods are higher than the de-trended average level. As time moves towards the tax hike, the coefficients grow both in absolute magnitude and in statistical significance. This pattern is present in both countries and no sizable difference between countries is observed. However, the coefficients for periods after the tax hike show a striking dif-ference between Japan and Korea. Coefficients of these periods for Korea are still significantly posi-tive, showing an uninterrupted continuation of the

pattern observed in periods preceding Japan’s tax hike. This observation is reasonable and expected since the output level in Korea is generally consid-ered immune to domestic policy shocks in Japan. The three coefficients for Japan, on the hand, all become statistically insignificant with p-values as high as 0.459, 0.934 and 0.742 respectively. The t-test results for the difference in coefficients shown in Table.9 confirms the above finding.

In Japan, the pattern observed before the VAT hike is noticeably disrupted by the hike. The dra-matic difference before and after the VAT hike im-plies that the production level in periods preceding the hike is significantly higher than that in the peri-ods immediately following the hike, which is con-sistent with the simulation result that the VAT hike lowers output level. The fact that production level in Japan during periods immediately before the tax hike is significantly higher than the de-trended av-erage might be an indication of forward-looking behavior by households and firms. As the time and magnitude of consumption tax hike become com-mon knowledge before the actual implementation, forward-looking households now have the incen-tives to purchase before the tax hike raises the cost of consumption. The collective behavior of theses households creates a temporally positive demand shock in periods immediately preceding the tax hike (a shifting of consumption from a future to the pres-ent). Sensitive to such shock, the profit-maximizing firms will respond by raising production level. The high level of production however, is only temporally that at the time of VAT hike, demand plummets and production returns to normal level. In reality, such fluctuation in demand is widely observed in Japan around the time of VAT hike especially in the market for durable goods such as motor vehicles. After revealing the distortionary effects on pro-duction as well as possible forward-looking behav-iors bought by the VAT hike, we now turn our focus to the demand side of the economy. The total retail trade is presumably more elastic to consumption tax shocks and therefore is expected to be more respon-sive to the VAT hike compared to production. Re-gression results of the impacts on total retail trades are summarized below in Table.10.

Page 22: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

22 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

The Breusch-Pagan test for heteroskedasticity performed on above regressions indicate consider-able problem of heteroskedasticity with p-value of 0.0147, hence robust standard errors are reported instead of normal errors. The results shown in Ta-ble.10 show that the effect of a one-time VAT hike on total retail trade is again negligible in Japan and Korea after controlling for other variables. The sig-nificance of its coefficient for Korea shown in the second column is biased as it is very small in magni-tude and disappears after the inclusion of more vari-ables. In consistency with the presumption about the elasticity of total retail trades to VAT hike, all coefficients for binary variables indicating periods surrounding the tax hike are statistically significant in Japan. More specifically, coefficients for periods before the tax hike are all significantly positive and those for periods after the tax hike are all signifi-cantly negative. This pattern resembles the impacts of VAT hike on production level but in a more dra-matic manner. In contrast, none of coefficients for periods preceding the tax hike are significant for Korea. The difference in coefficients between coun-tries is again confirmed by the t-test results shown in Table.11 below.

The striking difference in total retail trades be-fore and after the VAT hike in Japan indicates that the sales level in periods preceding the hike is sig-nificantly higher than that in the periods immediate-ly following the hike. This finding is also consistent with the simulation result that the VAT hike lowers consumption level. And the fact that the retail sales level in Japan during periods immediately before VAT hike is significantly higher than the de-trended average while that during periods immediately after the hike is significantly lower could be additional evidence in favor of the forward-looking behavior described above. Finally, we turn to our discussion on the trad-eoff between tax revenue and consumption distor-tion. Simulation results (Table. 6) showed that the initial VAT implementation at 3% and a VAT hike from 3% to 5% would boost tax revenue by 20.1% and 10.9% respectively with relatively little distor-tion on consumption (-0.96% and -0.63% respec-tively). In contrast, empirical evidence reveled that when Japan introduced VAT (at 3%) in 1989 tax revenue rose only by 8.3%, and following the VAT hike (from 3% to 5%) in 1997 tax revenue in turn contracted by 2.3% while private consumption fell by 3.6%. The discrepancy between simulated re-sults and empirical data can be primarily attributed to the act of exogenous shocks (notably the 1997 Asian Financial Crisis) that are not captured by the model. However, above analysis has shown that the distortion effect on Japanese consumption remains significant even when such shocks have been con-trolled for.

VI. Conclusion

Based on a static micro-founded General Equi-librium Model, the simulated consumption tax hike has negative impacts on utility, total output (con-sumption), labor supply and firm profits. And it has positive impacts on leisure and tax revenue. The mag-nitude of these impacts is proportional to the magni-tude of the tax hike though a diminishing marginal effect pattern is observed for consumption (total out-put) and tax revenue. All simulation results are ro-bust to the varying parameter values. Moreover, the magnitude of impacts on utility is particularly inelas-tic to the level of income endowment, implying that the welfare costs of VAT hike are similar to house-holds with different levels of income endowment. The regression analysis on empirical data from Japan and Korea during the periods from 1980 Q2 to 2013 Q4 substantiates the distortion effects of VAT hike on production and consumption. The magni-

Page 23: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 23

Volume xxiii. issue i.

tude of these effects in Japan is significant even af-ter other exogenous shocks have been controlled for. The striking contrast with Korea during the periods surrounding the VAT hike reveals the true partial ef-fects of the tax hike and indicates that its impacts on economy cannot be neglected. In addition, a unique pattern is observed in Japan during these periods. Both the production and retail sales level in periods immediately precede the tax hike is significantly higher than that in periods immediately follow as well as the de-trended average. This pattern might serve as indication of a forward-looking behavior by firms and households. Motivated by the incentive to avoid higher consumption costs after the VAT hike, forward-looking households rush to purchase dura-ble goods in periods preceding the tax hike, creating a positive demand shock and thus a temporary surge in production during these periods. Additionally, both of Japan’s past VAT hikes failed to achieve the theoretical level of tax revenue boost which points to a significantly large distortion effect of VAT hikes in Japan. In conclusion, the em-pirical evidence from Japan shows that the distor-tion impacts of VAT hike are consistent with simu-lation based on static General Equilibrium Model (GEM) and are significant in magnitude.

VII. References

Alm, J., & El-Ganainy, A. (2013). Value-Added Taxation and Consumption. International Tax and Public Finance, 20(1), 105-28.

Arnold, J. (2008). Do Tax Structures Affect Aggre-gate Economic Growth?: Empirical Evidence from a Panel of OECD Countries. OECD Economics De-partment, OECD Economics Department Working Papers.

Bhattarai, K. (2006). Macroeconomic impacts of consumption and income taxes: A general equilib-rium analysis. Indian Economic Journal, 54(2), 95-116.

Blanchard, O., & Fischer, S. (1989). Lectures on Macroeconomics. Cambridge, Mass: MIT Press.

Blundell, R. (2009). Symposium on the Econom-ics of VAT Cuts: Assessing the Temporary VAT Cut Policy in the UK. Fiscal Studies, 30(1), 31-38.

Bobenrieth H., E. S., Bobenrieth H., J. R., & Wright, B. D. (2008). A Foundation for the Solution of Con-sumption-Saving Behavior with a Borrowing Con-

straint and Unbounded Marginal Utility. Journal of Economic Dynamics and Control, 32(3), 695-708.

Boeters, S., Bohringer, C., Buttner, T., & Kraus, M. (2010). Economic Effects of VAT Reforms in Ger-many. Applied Economics, 42(16-18), 2165-82.

Bye, B., Strom, B., & Avitsland, T. (2012). Welfare Effects of VAT Reforms: A General Equilibrium Analysis. International Tax and Public Finance, 19(3), 368-92.

Caspersen, E., & Metcalf, G. (1995). Is A Value Added Tax Regressive? Annual Versus Lifetime In-cidence Measures.

Christandl, F., Fetchenhauer, D., & Hoelzl, E. (2011). Price Perception and Confirmation Bias in the Context of a VAT Increase. Journal of Economic Psychology, 32(1), 131-41.

Claus, I. (2013). Is the Value Added Tax a Useful Macroeconomic Stabilization Instrument? Econom-ic Modelling, 30, 366-74.

Crossley, T. F., Low, H., & Wakefield, M. (2009). Symposium on the Economics of VAT Cuts: The Economics of a Temporary VAT Cut. Fiscal Studies, 30(1), 3-16.

Fernandez-de-Cordoba, G., & Torres, J. L. (2011). The Transitory VAT Cut in the UK: A Dynamic General Equilibrium Analysis. Economic Issues, 16(1), 1-18.

Keen, M. (2007). VAT Attacks! International Tax and Public Finance, 14(2), 365-81.

Keen, M. M. (2011). Raising the consumption tax in Japan: Why, when, how? Washington, D.C.: Inter-national Monetary Fund.

Keen, M., & Lockwood, B. (2010). The value added tax: Its causes and consequences. Journal of Devel-opment Economics, 92(2), 138-151.

Lang, M. M. (2009). Value added tax and direct taxation: Similarities and differences. Amsterdam: IBFD .

Manente, M., & Zanette, M. (2010). Macroeconom-ic Effects of a VAT Reduction in the Italian Hotels and Restaurants Industry. Economic Systems Re-search, 22(4), 407-25.

Page 24: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

24 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

Miki, B. (2011). The effect of the VAT rate change on aggregate consumption and economic growth. New York: Center on Japanese Economy and Busi-ness, Columbia Business School.

Milesi-Ferretti, G. M., & Roubini, N. (1998). Growth Effects of Income and Consumption Taxes. Journal of Money, Credit, and Banking, 30(4), 721-44.

Ministry of Strategy and Finance, K. (2012). Ko-rean Taxation. Retrieved from Ministry of Strategy and Finance, Republic of Korea : http://www.nts.go.kr/eng/data/KOREANTAXATION2012.pdf

Narayan, P. K. (2003). The Macroeconomic Impact of the IMF Recommended VAT Policy for the Fiji Economy: Evidence from a CGE Model. Review of Urban and Regional Development Studies, 15(3), 226-37.

Nishiyama, Y. (2012). Main Issues for a Good Val-ue Added Tax System. Public Policy Review, 8(5), 683-703.

OECD. (2013). Main Economic Indicators - com-plete database. Retrieved from OECD: http://dx.doi.org/10.1787/data-00052-en

Rao, S. T. (2013). Impact of Value Added Tax (VAT) on Product Market Prices--A Study in State of Andhra Pradesh. International Journal of Research in Commerce and Management, 4(4), 125-29.

Syed, M. H., Tokuoka, K., & Kang, K. (2009). ‘Lost Decade’ in Translation: What Japan’s Crisis Could Portend about Recovery from the Great Recession. Washington D.C.: International Monetary Fund.

Tait, A. A. (1988). Value-added tax: International practice and problems. Washington, D.C. : Interna-tional Monetary Fund.

Wooldridge, J. M. (2013). Introductory Economet-rics: A Modern Approach, 5th Edition. Mason, OH: Cengage Learning.

VIII. Appendix

Appendix. A

Appendix. B

Page 25: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 25

Volume xxiii. issue i.

APPENDIX. C

APPENDIX.D

Page 26: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

26 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

Nahiomy AlvarezWilliams College

When National and Local Policies Clash: How Seattle’s Increase in Minimum Wage

Could Affect EITC Eligibility

I. Introduction

National and local policies often interact in ways that supplement or reinforce each other. For instance, to help low-income families, the federal government offers the refundable Earned Income Tax Credit, which, as of 2014, was supplemented in twenty-six states by a state-based EITC. In turn, the value of both the federal EITC and a state EITC supplement to a low-wage worker can depend on minimum wage policy. Minimum wage policy, like the EITC, is also regulated at a national and local level. Federal legislation imposes a uniform nation-al minimum wage (currently $7.25 per hour), which many local governments enhance with a higher hourly wage mandate. President Obama’s 2015 Fiscal Year Budget includes a series of proposals that would double the amount of EITC benefits for childless workers and raise the current federal mini-mum wage to $10.10 per hour for all workers.

Existing economic research on the EITC and minimum wage has centered around the employ-ment effects of the large expansions of the federal EITC and the relatively small increases in the fed-eral minimum wage. While the federal EITC may be substantially larger than a state’s EITC, the re-verse is true for the minimum wage. By imposing higher local minimum wages, states and cities can more drastically affect low-wage workers. On June 2nd, 2014, the City Council of Seattle, Washington, passed legislation to raise the minimum wage from its current $9.32 per hour to $15 per hour beginning April 2015, making it the highest minimum wage in the county. Now that the impact of Seattle’s mini-mum wage policy is being closely monitored -- al-ready influencing a national debate that, for years, has centered on unemployment effects -- the inter-action of a significantly higher minimum wage with other antipoverty policies merits careful consider-ation. Specifically, in gauging the comprehensive

EditorialThe minimum wage has been a central topic of discussion in politics and economics for quite some time. But a key issue rarely found in the headlines has to do with the conflict of local minimum wage policies and their affect on workers’ eligibility for federal tax credit programs. Recently the city of Seattle raised its minimum wage to the highest in the United States, providing Nahiomy Alvarez with a unique case study to examine this clash. Nahiomy Alvarez’s paper takes us through how the increase in Seattle’s minimum wage will affect workers’ eligibility for the federal earned income tax credit (EITC), one of the largest federal ‘anti-poverty’ programs and its implications for labor supply decisions.

Kristina Hurley

In this paper, I analyze how a substantial increase in Seattle’s minimum wage – now the highest in the US— will affect workers’ eligibility for the EITC. I find that the benefit of a $15 per hour minimum wage is partially offset by the resulting EITC changes. That is, despite generous EITC parameters, I estimate that the higher minimum wage will shift 36 percent of EITC recipients in Seattle out of eligibility, including 10 percent of whom were originally in the phase-in and plateau regions of the EITC schedule. I also provide the demographic characteristics of individuals who were shifted around and out of eligibility and calculate the net gains of all EITC recipients. I find that after the minimum wage increase, among those who were not shifted out of eligibility, the credit amount was unchanged for 16.7 percent, higher for 31.6 percent and low-er for 51.6 percent of recipients. Shifts in EITC eligibility are important because workers in different regions of the EITC face different effective marginal tax rates, which in turn can influence labor supply decisions.

Page 27: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 27

Volume xxiii. issue i.

impact of an increased minimum wage, we ought to examine if, and the extent to which, this increase in wages leaves workers ineligible for other forms of public assistance. In this paper, using 2011 American Community Survey data from Seattle, I predict how the increase in the earnings of low-income workers will affect eligibility for the EITC. Because families with an annual income as high as $43,998 may meet EITC eligibility criteria, it is possible that the increase in earnings of low-income workers will shift cer-tain workers from the phase-in region of the EITC schedule to the plateau or phase-out region. Such shifts are important because the effective marginal tax rates EITC recipients face vary by EITC region, which can thereby influence labor supply decisions. My analysis of Seattle suggests that, despite the generous EITC parameters, a $15 per hour mini-mum wage will shift 36 percent of EITC recipients out of eligibility. Thus, for at least 35 percent of the EITC population, the benefit of a higher mini-mum wage is partially offset by the resulting EITC changes. Although we might expect that most work-ers who were shifted out of eligibility were origi-nally in the phase-out region, 10 percent of EITC recipients were, in fact, originally in the phase-in and stationary regions. Additionally, 15 percent of those who were originally in the phase-in region will be moved to the phase-out region. I find that, among the EITC workers who were shifted out, over half were women, and over half had annual in-comes that put them below the poverty line. Among those who were not shifted out of eligibility, after the wage increase, 16.7 percent had the same credit amount, 31.6 percent had a higher credit amount, and 51.6 percent had a lower credit amount. My results suggest that increasing the minimum wage to $15 per hour in Seattle has significant conse-quences on EITC eligibility, and on the net gains of EITC recipients. While these results apply only to Seattle, public discourse and policymakers na-tionwide must consider this comprehensive analy-sis, rather than continuing to evaluate the minimum wage solely in terms of unemployment effects.

II. Existing Literature

2.1 Minimum Wage For many decades, economists have researched the effects of the minimum wage on labor market outcomes. The wide range of estimates, over the past decade alone, has made it increasingly diffi-cult to predict the effects of raising the minimum

wage. On the one hand, the leading conservative voices, David Neumark and William Wascher, posit that, while a higher minimum wage will help boost earnings, it will result in fewer jobs as employers are forced to pay higher wages (2008). According to this view, the benefits of raising the incomes of a few come at the expense of low-skilled jobs for many. On the other hand, Card and Krueger (1994) contend that with a small raise in the minimum wage, there are no significant adverse employment effects. Recent economic debate and research has focused on the minimum wage’s effects on employ-ment and hours worked, in part because it has been found the effect of the minimum wage on poverty is small. For instance, a study examining how a higher minimum wage affects single mothers specifically, finds that the federal minimum wage increases be-tween 1988 and 2003 had no effect on state poverty rates, among all poor or even among single mothers alone (Sabia & Burkhauser 2007). Another study (not focusing on single mothers) finds that over half of the total benefits of a higher minimum wage go to families that already earn wages above the poverty line (Formby et al. 2010).

2.2 The EITC Unlike traditional welfare programs, such as AFDC and Food Stamps, the EITC incentivises la-bor force participation: benefits increase with fam-ily earnings before eventually phasing out at higher incomes (Hoffman 2003). As Figure 1 shows, there are three distinct phases that EITC beneficiaries might fall in: the phase-in region, the stationary phase (or plateau), and the phase-out region. Which region each tax unit (an individual or family) falls into, will vary based on marital (filing) status, num-ber of qualifying children, and the amount of earned

Page 28: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

28 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

income. A tax filer without children must be at least 25 years old and under 65 years of age to claim the EITC. Nationally, 19 to 20 percent of all filers claim this tax credit. In terms of its effects on poverty, compared to the minimum wage, the EITC is a better-targeted anti-poverty tool. In 2001, exploiting state-level varia-tion in the EITC, Neumark and Wascher found that the EITC helps families rise above poverty-level earnings, primarily by inducing labor market en-try in families that initially did not have an adult worker. Throughout the next decade, Sabia and Burkhauser (2007, 2008, 2010, 2012) published several papers showing the small effects of mini-mum wage on poverty rates, further supporting the superiority of the EITC as an antipoverty program. Due to the EITC’s success, at least 27 states (includ-ing Washington State) now have an Earned Income Credit that supplements the federal EITC, although they tend to be relatively small. While there is still much disagreement among labor economists regarding the unemployment ef-fects and the behavioral responses to the minimum wage, there is broad consensus that the EITC has increased employment rates substantially. Meyer and Rosenbaum (2001) estimate the expansion of the EITC during the early 1990s increased the em-ployment of single mothers by 7 percent.

2.3 Interactions of EITC and the Minimum Wage Most of the research that has been conducted thus far has either compared the EITC and the mini-mum wage, or it has studied them independently; few studies have actually examined their interac-tion. One recent study, however, examined the ef-fect of a simulated 10% and then a 25% increase in the minimum wage and whether it enhances the effectiveness of the EITC. It finds that a higher minimum wage interacts with the current EITC in a way that both amplifies the labor supply response and increases the earnings of single women with children, but has adverse effects on the employment and earnings of low-skilled and minority individu-als without children (Neumark & Wascher 2011). These effects are exacerbated when a higher mini-mum wage is coupled with a high EITC. No studies thus far have examined how raising the minimum wage affects eligibility for the EITC, per se. As pub-lic discourse about raising the minimum wage in-creases, its interaction with the existing EITC ought to be considered.

III. Seattle

Seattle, like many other cities, is currently fac-ing a growing level of income (and wage) disparity: over 50 percent of all earnings currently go to the top quintile, while the rest is distributed throughout the rest of the population. To address this issue and as an attempt at reducing poverty rates among mi-norities and low-skilled workers, city council mem-bers have recently passed Ordinance No. 124490, which mandates an increase in the minimum wage from its current $9.32 per hour to $15 beginning in April of 2015. Seattle’s new minimum wage will be implemented steadily over the next several years by local businesses at a pace dependent on size and benefits provided by the employer. Large employers (with more than 500 employees) that do not provide their employees with a health plan are the first that will face the $15 per hour minimum, which they are required to fully implement by 2017. One way to determine the effectiveness of this policy is to first examine who will be affected by this policy change. Determining whether a signifi-cant increase in minimum wage is an effective an-tipoverty tool, ultimately, also depends on how this policy interacts with at least one other major anti-poverty policy, the EITC.

IV. Methods

To estimate how the proposed increase in mini-mum wage to $15 per hour will affect EITC eligibil-ity for low-wage workers in Seattle, I used IPUMS data drawn from the 2011 American Community Survey. IPUMS provides extensive demographic and labor market information for a representative sample of 5,604 non-institutionalized individuals and their families living in Seattle. I was particular-ly interested in looking at the characteristics of low-wage workers and individuals eligible for the EITC, for which I gathered information on income, hours worked, family structure, education, age, race, sex, and poverty status . Because the proposed law in Seattle does not go into effect until 2015 and will be implemented in phases, I made a forward-looking analysis of how the increase in the minimum wage will affect these individuals by 2018, the first year when large employers are mandated to fully imple-ment the higher wage and when small employers are mandated to guarantee minimum compensation. Individuals surveyed by the American Commu-nity Survey do not directly report if they are paid the minimum wage or the precise number of weeks worked. Hence, I calculated hourly wages by divid-

Page 29: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 29

Volume xxiii. issue i.

ing the total annual earnings of each worker by the total annual hours worked. To obtain total annual hours worked, I multiplied the number of hours per week that the respondent usually worked (if the per-son worked during the previous year) by the num-ber of weeks worked in the previous year. Weeks worked was originally given in intervals, so I first assigned each worker the number of weeks equal to the midpoint of his interval. To determine eligibility for the EITC, I added the earnings of all family members including self-employment income, determined the number of children and the marital status of the household head, whether they lived in group quarters, and their citizenship status. If families met EITC eligi-bility criteria, their income placed them in either the phase-in, plateau, or phase-out of the EITC. To estimate the effects of the minimum wage increase on EITC eligibility, I first identified those currently earning less than $15 per hour. I estimated how the increase would affect their earnings by rais-ing their hourly wage to $15 and multiplying the new wage by the total number of hours each person worked in 2011. I made the assumption that workers whose wage was near the proposed minimum wage were unaffected by the hypothetical increase in the minimum wage, and that no changes in employment or hours would result from the higher minimum wage. This is a simplifying assumption that can be relaxed in future research. I then used the workers’ new annual wage income to calculate the worker’s family’s total income andto determine whether the new total income would shift them on the EITC schedule. I also calculated net gains for recipients who were not shifted out. Finally, I determined whether the family was in poverty before and after the minimum wage in-crease. Current poverty status is available directly in the ACS data for 2011. To verify that my calcu-lations of poverty status after the minimum wage increase were reliable, I used the same methodology to impute current poverty status for 2011. There was tight correspondence between my imputed current poverty variable and the Census poverty indicator, with a correlation of .98. In all my tables, poverty ratio refers to the ratio of family income to the of-ficial poverty threshold for a family of a given size and composition, with 2, for instance, representing incomes at 200% of the poverty level. Before turning to my findings, it is important to understand the limitations of the ACS data for this purpose and how I tried to compensate for those problems. First, the ACS does not ask directly

whether the family claimed the EITC. It does, how-ever, provide information for each household on the key parameters of the EITC eligibility formula: the amount of earned income, number of children, marital status, citizenship status, and age, which I can use to determine EITC eligibility. I make the assumption that those who are eligible for the EITC claim it. Nonetheless, the accuracy of esti-mates derived from ACS data depends on how the information taxpayers reported on their tax forms differs from the data reported to the ACS. After determining eligibility using IPUMS data, I com-pared it to the information published by the IRS on EITC receipt in 2011. I found that the two sets of estimates are similar. Using IRS data on federal in-dividual income tax filers provided by the Brook-ings Institute, I calculated that about 10.1 percent of tax filers in Seattle claimed the EITC. My own data shows that 12.4 percent of all Seattle residents were eligible for the EITC in the same year. There are numerous reasons the number of tax units that claimed the EITC could be lower than the number that is eligible. First, not everyone who is eligible will claim the EITC. For instance, studies show that taxpayers with no qualifying children have lower rates of participation than taxpayers with qualifying children (Plueger 2009). Secondly, people might be reluctant to share their citizenship status in the cen-sus, and it is possible that some of the people who I determine are eligible for the credit might in fact be ineligible – something that the IRS data would be more precise in reporting. Thirdly, when reporting income to the Census Bureau, it is possible that a re-spondent could provide rounded income (unless the respondent has tax return documents in hand). In cases where the taxpayer is very close to the EITC thresholds, this could influence my estimates.

V. Description of Data

Table 1 reports descriptive statistics for sex, age, race, education, family structure, poverty sta-tus, hours and weeks worked for all Seattle work-ers. It shows how workers within each demographic category, say all women, are distributed across four wage categories. The four wage categories are di-vided as follows: the current Seattle minimum wage of $9.32 per hour or less, between $9.32 and $10.10 per hour (the higher federal minimum wage recently proposed by President Obama), between $10.10 and $15 per hour (Seattle’s proposed minimum wage), and over $15 per hour. According to my estimates, out of all workers in Seattle, 16.4 percent earned

Page 30: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

30 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

the current minimum wage ($9.32 per hour or less), and another 16.8 percent earned $15 per hour or less (Table 1A). Put together, this means at least one in three workers could be affected by the proposed in-crease. At the current state minimum wage of $9.32 per hour, a single parent in Seattle, with 2 children, working 40 hours/week, all 52 weeks of the year (no breaks), earns $19,386 before taxes – putting them barely above the poverty line. According to city reports, in 2011, only 15 percent of all Seattle’s residents had incomes below the official poverty line. However, the same report by the University of Washington shows that low-income workers make up the largest group of those currently earning the minimum wage: 40 percent of such workers live in families at or below the poverty threshold, and an-other 27 percent live in families with incomes 200 percent of the poverty threshold or less. Figure 2 illustrates the poverty demographic group in Table 1, and how poor workers are dis-tributed across the four wage categories. I find that among all poor workers, over two thirds earned the current minimum wage and only 13.2 percent of all poor workers earned over $15 per hour. Table 2 reports the same demographic groups shown in Table 1, but instead, the table shows how workers within each of the wage categories are distributed across demographic groups. For instance, out of all workers earning the minimum wage, 48.2 percent worked less than 20 hours per week, 24.6 percent worked between 20 and 34 hours, and 27.2 percent worked full-time. And, out of all workers earning the minimum wage, women made up a larger por-tion of workers earning the current minimum wage or between $9.32 per hour and $10.10 per hour.

The University of Washington study also finds that, although non-Hispanic whites are a majority of all minimum wage workers in Seattle, women and minorities are disproportionately represented among workers earning the minimum wage. I find

that well over 40% of Black, Asian/Pacific Islander, and Hispanic workers, and American Indian/Alas-kan Native workers earn less than $15 per hour, compared to about 29 percent of non-Hispanic white workers (Table 2).Table 1. Percentage of all Earners in Each Wage Category by Demographic Characteristic, 2011 (n=3290)

Table 2. Percentage of Seattle’s Low Wage Workers within Each Demographic Group, 2011

In 2011, 12.4 percent of Seattle’s population met eligibility requirements for the EITC. Given the EITC’s generous parameters, it is

Page 31: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 31

Volume xxiii. issue i.

unlikely that everyone who is EITC-eligible is also a minimum wage worker. However, the EITC is an anti-poverty program that primarily targets low-income families, and given that –as Figure 2 shows— a large fraction of low-income workers will be affected by the new minimum wage policy, the overlap between the minimum wage and EITC populations merits consideration. My estimates show that among all EITC-eligible workers, over half earned the minimum wage, and almost 80 per-cent earned below $15 per hour. Moreover, because a higher minimum wage directly impacts earnings, we can assume tax units who were previously eli-gible for the EITC may see a change or loss in their tax credit, after the minimum wage raises their in-come. Of particular importance are those individu-als in the phase-out region who could be facing the highest marginal tax rates and losing EITC benefits altogether. Among all EITC-eligible workers earn-ing the minimum wage in 2011, 46.7 percent were in the phase-in region, 16 percent in the plateau, and 36.7 in the phase-out region; and, that over two thirds all workers earning between $9.32 per hour and $10.10 per hour were in the phase out region. When including children in the estimates, the pro-portion of EITC recipients in the phase out region increases from the aforementioned 36.7 percent to 45.2 percent. How this distribution changes after the minimum wage legislation is crucial when con-sidering net gains and losses.

VI. Results

Given the way EITC recipients are distributed on the EITC schedule, I expect that many work-ers who were income-eligible for the EITC in 2011 will no longer be income-eligible in 2018. My esti-mates support this hypothesis: I find that 35 percent of the families who were eligible for the EITC in 2011 will be shifted out of eligibility after the mini-mum wage increase. While in 2011, 12.4 percent of Seattle’s population was eligible for the EITC, after the law is implemented only 8 percent will be income-eligible. Figure 2 shows that after the raise, about 36 percent of all EITC recipients will be in the phase-in, down from 42 percent in 2011; about 9 percent will be in the plateau, down from 12 percent in 2011; and 55 percent will be in the phase-out, up from 46 percent in 2011.

Figure 3: EITC Recipient Distribution Before and After Minimum Wage Raise

Though this change in distribution is relevant because of different marginal tax rates and chang-ing credit amounts, from a policy perspective the people for whom the policy is most important are the people who lose eligibility after the raise. I find that out of those who were originally in the phase in, 14.7 percent are shifted out of eligibility; out of those who were originally in the plateau, 32.1 per-cent out of eligibility; and, more than half of those who were originally in the phase-out are shifted out of eligibility (Table 3). Additionally, I find that the increase in wages is predicted to reduce the poverty rate among EITC recipients from 53.7 percent to 42.8 percent assuming employment and hours did not change.Table 3. Percentage of EITC Recipients in Each EITC Phase Before and After the Minimum Wage Increase

Finally, I calculate the net gains of EITC recipi-ents not shifted out of eligibility to have a general idea of whether Seattle’s low-wage workers who did not lose eligibility were indeed better off after the minimum wage increase. But first, my calcula-tion of net gains does not take into consideration Washington’s state EITC, and secondly, because the rounding of income by an ACS survey respon-dent could impact the estimated dollar amount of the credit, my calculations could either overstate or understate the net gains/loss of EITC recipients. With that in mind, I find that among those who were not shifted out of eligibility, 16.7 percent had the same credit amount, 31.6 percent had a higher credit amount after the raise, and 51.6 percent had a higher credit amount before the raise. I estimate that, of those who are shifted out of eligibility, over 80 per-cent are minimum wage workers.

Page 32: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

32 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

VII. Conclusion In this paper, I examined the interaction between the proposed minimum wage and the Earned In-come Tax Credit. I find that, for a significant por-tion of Seattle’s population, the benefits of a higher minimum wage are partially offset by the resulting change in EITC eligibility. Thousands of Seattle’s low-income residents currently benefit from other forms of federal and state assistance, such as Sec-tion 8 Housing, Food Stamps, National School Lunch Program, and utility assistance programs. With significantly higher wages, it is possible that some of Seattle’s low-income workers may no lon-ger be eligible for some. This paper’s findings have implications for whether programs in place are ef-fective or need restructuring, what groups need more assistance than others, and, ultimately, how to best help low-income workers. Until we weigh the impact of the significantly higher minimum wage in conjunction with other forms of assistance, we cannot unequivocally conclude that it will be an ef-fective and well-targeted antipoverty tool. Moreover, the variation in the use of the ben-efits from these different antipoverty policies is key to understanding why poor and low-wage workers need both a minimum wage that allows them to meet the cost of living, and an EITC to al-leviate their financial burdens annually. While the minimum wage provides a small wage boost each month to ameliorate rising costs of food, utilities, and basic needs, the annual lump-sum credit helps low-income and even moderate-income workers pay off bills or debt, pay for more expensive (du-rable) goods and services, and could allow them to save, thereby improving financial stability (Barrow and McGranahan, 2000; and Smeeding, Phillips, O’Connor, 2000). When the benefits of increasing the minimum wage work to offset the benefits of another antipoverty place, this policy is not the most efficient way to help low-income workers. Finally, some may argue that, while the EITC is a more effective antipoverty tool than the minimum wage in the long run, in the short run, low-income families, especially those living below the poverty line and struggling to meet basic needs, benefit more from a raise in the minimum wage. In this study, I find that approximately one-third of indi-viduals previously eligible for the EITC will no lon-ger be able to receive it, a direct cost of the higher minimum wage. However, there may be other costs that have yet to be researched, or that could also be much harder to quantify. For instance, to offset

the higher costs of the minimum wage increase, em-ployers may choose to terminate fringe benefits (re-tirement plans, health insurance, paid holiday and vacation days), or limit amenities such as free meals and parking. The subsequent cost of obtaining these goods individually cannot be discounted. Again, careful consideration must be given to ensure that antipoverty programs complement each other. Notably, this forward-looking analysis that I have conducted is static. It does not account for the possibility that firms may increase the wages of their other employees already earning near the proposed minimum wage; that the composition of the work-force may change and demand for certain skill-level might increase; or that businesses may relocate or reduce the size of their workforce or the number of hours demanded. However, many of these assump-tions can be relaxed, opening the field for future research. It could also be of value to conduct quali-tative research after the law is implemented to de-termine whether individuals’ subjective responses coincide with the existing empirical research.

VIII. References

Barrow, Lisa, and Leslie McGranahan. (2002). “The Effects of the Earned Income Tax Credit on the Seasonality of Household Expenditures,” Na-tional Tax Journal, Vol. 53, No.4, part 2, December, (2000): 1211-1244.

Brown, Charles, and James Medoff. “The Employer Size-Wage Effect.” Journal of Political Economy 97, no. 5 (1989): 1027-59.

Bureau of Census, Income, Poverty, and Health In-surance Coverage in the United States: 2011, Cur-rent Population Reports, P60-229 (August 2011).Card, David; Krueger, Alan B. (September 1994). “Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Penn-sylvania”. The American Economic Review 84 (4): 772–93.

Eissa, Nada, and Jeffrey B Liebman. “Labor Supply Response to the Earned Income Tax Credit.” Quar-terly Journal of Economics, Vol. 85, No.2 (1995): pp. 394-408.

Formby, John P., John A. Bishop, and Hoseong Kim. “The Redistributive Effects and Cost-Effectiveness of Increasing the Federal Minimum Wage.” Public Finance Review 38, no. 5 (2010): 585-618.

Page 33: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 33

Volume xxiii. issue i.

Forward Seattle. www.forwardseattle.org

Hoffman, Saul D., and Laurence S. Seidman. Help-ing Working Families: The Earned Income Tax Credit. Kalamazoo, Mich.:W. E. Upjohn Institute for Employment Research, 2003.

Keenan, Nicole, and Howard Greenwich. “Equity Outcomes of a Economic and Equity Outcomes of a $15/Hr Minimum Wage in Seattle.” Puget Sound Sage, 2014.

Klawitter, Marieka M., Mark C. Long, and Robert D. Plotnick. “Who Would Be Affected by an In-crease in Seattle’s Minimum Wage?” University of Washington: Evans School of Public Affairs, 2014.

Leigh, Andrew. “Who Benefits from the Earned Income Tax Credit? Incidence among Recipients, Coworkers and Firms.” B.E. Journal of Economic Analysis and Policy: Advances in Economic Analy-sis and Policy 10, no. 1 (2010).

Malul, Miki, and Israel Luski. “The Optimal Policy Combination of the Minimum Wage and the Earned Income Tax Credit.” B.E. Journal of Economic Analysis and Policy: Contributions to Economic Analysis and Policy 9, no. 1 (2009).

Martinez, Amy. “Tips Are Top Concern for Some in $15 Minimum-Wage Debate.” The Seattle Times, April 20, 2014.

Meyer, Bruce D., and Dan T. Rosenbaum. “Welfare, the Earned Income Tax Credit, and the Labor Sup-ply of Single Mothers.” Quarterly Journal of Eco-nomics 116, no. 3 (2001): 1063-114.

Neumark, David, and William Wascher. “Does a Higher Minimum Wage Enhance the Effectiveness of the Earned Income Tax Credit?” Industrial and Labor Relations Review 64, no. 4 (2011): 712-46.

Neumark, David, and William L. Wascher. Mini-mum Wages. Cambridge and London:MIT Press, 2008.

Opportunity for All: The President’s Fiscal Year 2015 Budget. White House: Office of Management and Budget. http://www.whitehouse.gov/omb/bud-get?_ga=1.205873424.2008938182.1407677068

Plueger, Dean. “Earned Income Tax Credit Partici-pation Rate for Tax Year 2005.” IRS Research Bul-letin (2009).

Reich, Michael, Ken Jacobs, and Annette Bern-hardt. “Local Minimum Wage Laws: Impacts on Workers, Families and Businesses.” University of California, Berkeley, 2014.

Rothstein, Jesse. “Is the Eitc as Good as an Nit? Conditional Cash Transfers and Tax Incidence.” American Economic Journal: Economic Policy 2, no. 1 (2010): 177-208.

Sabia, Joseph J., and Richard V. Burkhauser. “The Effectiveness of Minimum-Wage Increases in Re-ducing Poverty: Past, Present, and Future.” Con-temporary Economic Policy 25, no. 2 (2007): 262-81.

Sabia, Joseph J. “Minimum Wages and the Eco-nomic Well-Being of Single Mothers.” Journal of Policy Analysis and Management 27, no. 4 (2008): 848-66.

Sabia, Joseph J., and Richard V. Burkhauser. “Mini-mum Wages and Poverty: Will a $9.50 Federal Minimum Wage Really Help the Working Poor?”. Southern Economic Journal 76, no. 3 (2010): 592-623.

Sabia, Joseph J., Richard V. Burkhauser, and Ben-jamin Hansen. “Are the Effects of Minimum Wage Increases Always Small? New Evidence from a Case Study of New York State.” Industrial and La-bor Relations Review 65, no. 2 (2012): 350-76.

Smeeding, Timothy M., Katherin Ross Phillips, and Michael O’Connor. “The EITC: Expectation, Knowledge, Use, and Economic and Social Mobil-ity,” National Tax Journal, Vol. 53, No. 4, part 2, December, (2000): pp. 1187-1210.

Steven Ruggles, J. Trent Alexander, Katie Genadek, Ronald Goeken, Matthew B. Schroeder, and Mat-thew Sobek. Integrated Public Use Microdata Se-ries: Version 5.0 [Machine-readable database]. Min-neapolis: University of Minnesota, 2010.

$15 Minimum Wage. City of Seattle: Office of the Mayor. Retrieved July 30, 2014 from http://murray.seattle.gov/minimumwage.

Page 34: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

34 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

U.S. Department of Labor: Wage and Hour Divi-sion. www.dol.gov.

IV. Appendix

Table 6. Demographic Characteristics for 2011 EITC-Eligible Individuals Shifted Out of Eligibility

Table 2. Earned Income Tax Credit Parameters for Unmarried Filers, 2011

Under the 2011 EITC schedule, the credit C for a household with two or more children with labor earnings E and total income Y is given by:C=0.40E if E < $12,780C= $5,112 if E < $12,780 and Y < $16,690C= $5,112 – 0.2106 (Y– $16,690) if E < $12,780 and $16,690 < Y < $40,964C = 0 if Y > $40,964Source: Hoffman, 2003

Page 35: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 35

Volume xxiii. issue i.

I. Introduction

“Singapore’s exports slump amid Chinese slow-down!” screams the headline of the Straits Times, Singapore’s leading newspaper (Straits Times, 2014). When economic powerhouse China— Asia’s rising dragon—adjusts its consumption patterns, the region’s smaller nations must adapt quickly or suffer the consequences. Singapore, one of the four ‘Asian tigers’ known for rapid economic growth since the 1960s, has enjoyed commendable success in doing the former. It has relied on accurately forecasting the caprices of foreign demand, and gearing the sec-tors of its small, open and export-oriented economy to best meet these evolving needs. One-third of the island nation’s trade consists of oil and oil products, making it particularly vulnerable to fluctuations in the external demand for this natural resource. China alone consumes over a fifth of these oil exports, and changes in Chinese oil consumption patterns could have a significant impact on Singapore’s economy. While several studies have investigated the impact of China’s oil consumption on its own economy, no such cross-border effect has been explored. An increase in China’s demand for oil has two distinct and conflicting effects on Singapore’s oil exports. Firstly, it leads to a direct increase in the demand for oil exports, potentially boosting Sin-gapore’s oil sector. However, increases in China’s oil consumption are also paralleled by a growth in the Chinese oil refining industry, resulting in a de-crease in regional demand for Singapore’s refined

Pratyusha MukherjeePrinceton University

Crouching Tiger, Rising Dragon: The Effect of China’s Oil Consumption on

Singapore’s Economy

EditorialIn this illuminating piece, Pratyusha Mukherjee takes a look at Singapore, and explores how the tiny city-state was able acquire an outsize influence in today’s world. Singapore has managed to creatively plan the direction of its economy, helping the nation navigate the ever changing international economic landscape. Efforts to market the city as the gateway to Asia have been successful, and the city is truly an important player in the global economy. This thoughtful research into the Singaporean economy will be informative to any reader.

Jacob Miller

petroleum products. Singapore’s long-term eco-nomic strategies depend heavily on the domination of one of these effects over the other. A net posi-tive impact on Singapore’s economy may promote greater investment in the nation’s oil refining sec-tor, due to the persistent trend of increasing Chinese oil consumption. On the other hand, a detrimental effect would encourage Singapore’s diversification into other sectors. Figure 1 below summarizes the nature of the relationships between these four main variables of interest: Chinese oil consumption, Chi-nese refining capacity, Singapore’s gross domestic product (GDP) and Singapore’s merchandise trade.

Figure 1: Flowchart representing theoretical chan-nels of effect between Chinese oil consumption, Sin-gapore’s trade and Singapore’s GDP

To establish the directionality of this relation-ship, controlling for other variables that would otherwise lead to omitted variable bias, I measure

Page 36: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

36 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

the effect of China’s oil consumption on two key economic indicators of Singapore: real GDP and value of merchandise trade. Furthermore, due to the disproportionately large impact of the oil refining sector on Singapore’s trade, I hypothesize that Chi-nese oil consumption will lead to an increase in the merchandise trade of Singapore, but will not have a significant impact on Singapore’s more diversified GDP. In this paper, I find that Chinese oil consump-tion is positively correlated with Singapore’s trade, but does not translate into any measurable corre-lation with Singapore’s GDP. I then use the Toda-Yamamoto procedure for testing Granger-causality to establish that rising Chinese oil consumption Granger-causes an increase in Singapore’s merchan-dise trade, but that there is no causal relationship between China’s oil consumption and Singapore’s GDP.

II. Literature ReviewThe trade of Singapore has fascinated econometri-cians ever since the city’s founding in 1819 (Ken, 1960). A country barely 247 square miles in size, a territory slightly smaller than that of New York City alone, it has evolved from an obscure trading outpost into one of the world’s most highly-ranked economies. Today, it boasts a real GDP per capita of more than US$60 000, one of the highest in the world. This spectacular growth has been driven by leveraging the only natural resource the country possesses—a strategic geographical location be-tween the Indian and Pacific oceans—to launch it-self as the region’s central trading hub. Openness to international trade has thus been the cornerstone of Singapore’s economic success, but this success is as fragile as it is impressive. The very fact that Singapore is the most open economy in the world, with a trade to GDP ratio of over 250 percent, makes it critical that the city-state remain aware of and re-sponsive to shifts in international demand patterns, so as to remain relevant in the global trade markets (Rajan, 2009). This is particularly true in the case of the coun-try’s oil sector. Despite having no hydrocarbon re-serves of its own, Singapore is one of the world’s top three oil trading and refining hubs. It imports crude oil from oil-producing countries, mainly in the Middle East, for refining purposes, and then re-exports the refined products, such as gasoline and distillate fuel oil, to mainly Asian markets. Of these, China accounts for almost a quarter of Singapore’s total refined oil exports, as summarized by Figure

2 below. The trading activity generated from this sector forms more than a third of Singapore’s total trade, while the resulting value added to the econo-my accounts for 5 percent of Singapore’s total GDP, or about US$14 billion per year.

Figure 2: Singapore’s Refined Oil Exports by Des-tination (U.S. Energy Information Administration, 2013) China is the world’s second largest consumer of oil, and recently overtook the United States to be-come the largest net importer of oil. As China is a major importer of Singapore’s refined oil products, as well as one of its key competitors in the regional oil refining industry, changes in China’s oil con-sumption patterns have a profound impact on both global markets at large, and the Singapore economy in particular (Zweig & Jianhai, 2005). The effect of China’s rising energy use on its own GDP has been thoroughly analyzed. In particular, empirical studies on the relationship between oil consumption and real GDP in China have suggested a positive bi-directional causality between the two variables (Zhang & Yang, 2013). However, the effect of this oil consumption on the economy of another country has not been directly explored in existing literature. Nevertheless, the 2013 Oil Market Report may provide one link of a potential causation chain be-tween Chinese oil consumption and the Singaporean economy. It pinpoints growing Chinese oil demand as a leading cause for increasing crude oil prices (In-ternational Energy Agency, 2013). Chinese oil con-sumption pushes up the world price of crude oil, which in turn affects Singapore’s economy. Empiri-cal investigations have indeed confirmed that an in-crease in crude oil prices leads to a significant nega-tive impact on Singapore’s GDP growth in the long run (Abeysinghe, 2001). This suggests a possible negative effect of increasing Chinese oil consump-tion on the Singapore economy through the mecha-nism of crude oil prices.

Page 37: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 37

Volume xxiii. issue i.

In  million  barrels  per  day  

Net    imports    

Figure 3: China’s Oil Production and Consumption, 1990-2013 (U.S. Energy Information Administra-tion, 2014)

However, the above-presented model does not take into account the impact of increased Chinese demand for Singapore’s refined oil products, or the competition posed by China’s own refining sec-tor. The Singapore government is optimistic about trends of increasing oil consumption in China. As Figure 3 above indicates, the growth in China’s do-mestic oil production has thus far been lagging be-hind the rapid increase in Chinese oil consumption, resulting in an expansion of net Chinese oil imports. Singapore authorities hope to upgrade and expand their existing oil refining capacity in order to capi-talize on this increasing Chinese demand for refined oil products from external sources. The Economic Development Board of Singa-pore reported that the government aims to maintain Singapore’s share of global refining capacity by continued investment into the oil refining sector in order to enable the country to enhance the growth of its oil trading activities by creating the critical volume of export-oriented refining throughput (Le-ong, 2013). The government hopes that increasing refined oil production, by upgrading existing assets and investing in new ones, will help tap into grow-ing Chinese demand for refined oil products, boost-ing Singapore’s exports. This seems to indicate that Singapore has not yet maximized its returns from investing in oil refining activities. Conversely, industry reports suggest that fur-ther expansion in Singapore’s refining capacity may not be advisable, as the recent increase in the oil refining capacity of other countries in the region,

particularly in that of China, could reduce demand for Singapore’s refined oil exports (Yep, 2013). In addition, the profusion of these new plants will put pressure on refining price margins, the high price differences between crude oil and refined products that have made the oil refining industry so lucrative in recent years. This may have a dampening effect on Singapore’s oil refining sector, and consequently, on its economy as a whole. This paper aims to investigate these contrast-ing claims by examining the specific effect of Chi-nese oil consumption on the Singapore economy. By using two independent variables, Singapore’s real GDP and merchandise trade, I hope to deter-mine whether an increase in Chinese oil consump-tion affects the Singaporean economy as a whole (as measured by real GDP), or merely has an impact on its trade figures (as measured by total merchandise trade). This analysis thus extends existing research on the effects of Chinese oil consumption beyond its own borders to better inform the policy decisions governing the development of Singapore’s econo-my.

III. Data

I sourced data on China’s annual oil consump-tion from 1991 to 2010 from the China Statistical Yearbook publications by the National Bureau of Statistics of China. This time period covers the span of four of China’s five-year plans for economic de-velopment, providing a comprehensive look into the evolution of the country’s energy consumption pat-terns. While the eighth economic plan (1991-1995) under Deng Xiaoping marked a period of increasing energy consumption and foreign trade to fuel eco-nomic growth, the most recent eleventh plan (2006-2010) focused on reducing China’s dependence on crude oil imports by diversifying its foreign energy sources and investing in renewables. This time span thus provides a good indication of the developing trends in Chinese oil consumption as the nation tries to balance economic and environmental sustainabil-ity. I then utilized data from the World Develop-ment Indicators collected by the World Bank to obtain annual figures for Singapore’s real GDP, at 2005 prices. The use of real GDP instead of nominal GDP eliminates the noise created by varying infla-tion rates, thus providing a suitable measure for the overall impact of changes in Chinese oil consump-tion on the Singapore economy. However, it must be noted that despite the high value of trade and trans-

Page 38: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

38 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

actions stemming from the oil sector, net income from the industry contributes to only 5 percent of Singapore’s GDP. Thus, it is possible that the results of a regression analysis may not be statistically sig-nificant enough to establish the directionality of the net economic impact on Singapore’s GDP. To overcome this limitation, I also examined the effect of Chinese oil consumption on the value of Singapore’s merchandise trade. As noted earlier, the Singapore government’s justification for contin-ued investment into the oil refining sector is to maxi-mize trade, not necessarily to maximize GDP, which is a broader measure. Merchandise trade is over 300 percent of Singapore’s GDP, and 30 percent of this consists of oil and oil products (Lim, 2013), which magnifies the impact of increasing Chinese oil con-sumption on Singapore’s economy. A 10 percent in-crease in refined oil exports to China corresponds to a 3 percent rise in Singapore’s merchandise trade, but translates to a mere 0.5 percent increase in Sin-gapore’s GDP. Thus, the results of a regression of Chinese oil consumption on Singapore’s trade are more likely to be statistically conclusive than those of a regression on Singapore’s GDP. Data on the value of Singapore’s total merchan-dise trade was obtained from World Development Indicators database, which provided the value of trade as a percentage of GDP. To facilitate empirical analysis, I converted these percentages to absolute figures by multiplying them by the values of Singa-pore’s real GDP at 2005 prices obtained earlier. Ta-ble 1 presents summary statistics on the independent variable and two dependent variables of interest.

Table 1: Summary Statistics for Time Series of OC, S_GDP and S_MT

Figure 4 below presents a scatterplot of these three variables by plotting Chinese oil consumption against Singapore’s GDP and Singapore’s merchan-dise trade. From this we can see that the absolute level of Singapore’s merchandise trade is signifi-cantly higher than that of its GDP.

Figure 4: Scatterplot of oil consumption in China against key indicators of Singapore’s economy The scatterplot for trade also more granularly displays the fluctuations in trend associated with Chinese oil consumption, with the slight dip in GDP near the end of the x-axis being magnified into a sizeable drop in merchandise trade. From both graphs, however, it seems clear that increasing oil consumption in China is positively correlated with higher trade and GDP in the Singapore. Nevertheless, we cannot determine a causal re-lationship from this preliminary examination of the data, because it may be the case that other external factors are driving an increase in both Chinese oil consumption and Singapore’s economy. Hence, in order to accurately measure the relationship between these variables and avoid omitted variable bias, it is necessary to control for potentially confounding factors. The first of these control variables would be the price of crude oil. By the law of demand, an in-crease in oil prices will lead to a fall in Chinese oil consumption and will also have a dampening effect on Singapore’s economy, as laid out in Section II. I used the West Texas Intermediate benchmark of oil prices for this variable, averaging the daily spot prices over a year to obtain an annual measure. Secondly, I included a variable for Chinese GDP, at 2005 prices, from the World Bank database. An increase in Chinese GDP drives an increase in China’s oil consumption, since a positive bi-direc-tional causality has been established between Chi-na’s GDP and its oil consumption (Zhang & Yang, 2013). At the same time, China is one of Singapore’s largest trade partners, so an increase in China’s in-come level would also boost Singapore’s GDP. To ensure completeness, a control variable for the

Page 39: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 39

Volume xxiii. issue i.

GDP of Singapore’s top three trade partners (ex-cluding China)—Malaysia, the United States and Indonesia, respectively—was also included (Di-rectorate General for Trade, 2013). The variable captures the effect of fluctuations in Singapore’s economic indicators due to changes in the income levels of its major trading partners. In the case of the trade partner control variable, GDP per capita values were necessary instead of GDP values in or-der to control for the relative size of each of these countries. To calculate these, I collected data on the real GDP values of each of these three countries at 2005 prices, as well as their population sizes, from the World Development Indicators. I then generated a new variable using the following formula:

T_GDP = US_GDP/US_popn + M_GDP/M_popn + I_GDP/I_popn

where US_GDP, M_GDP and I_GDP are the real GDP values in US dollars at 2005 prices of the Unit-ed States, Malaysia and Indonesia respectively, and US_popn, M_popn and I_popn are the population counts for each of the three countries. T_GDP thus provides a weighted average measure of the income levels of each of Singapore’s three major trading partners. Finally, data on China’s refining capacity, in terms of the number of barrels of refined oil pro-duced per calendar day, was collected from the International Energy Statistics database. The inclu-sion of this variable helps to determine the impact of competition from China’s own expanding refining activities on Singapore’s oil refining and re-export-ing industry. This analysis in itself would be criti-cal in informing policy decisions by the Singapore government with regard to the oil refining industry by providing information about its competitiveness in regional markets. Summary statistics for the four control variables are presented below in Table 2.

Table 2: Summary Statistics for Time Series of OP, C_GDP, T_GDP and CRC.

IV. Methodology

The first component of my data analysis is a di-rect observation of the relationship between Chinese oil consumption and Singapore’s GDP. In order to measure the effect accurately, the issue of omitted variable bias must be addressed. Omitted variable bias occurs when the omitted variable is correlated with an included regressor, and is also a determinant of the dependent variable. As discussed previously, oil price, Chinese GDP, the GDP of Singapore’s ma-jor trading partners and Chinese refining capacity are all important omitted variables that will have to be accounted for in the regression in order to isolate the effect of increased Chinese oil consumption on Singapore’s GDP. The ordinary least squares (OLS) regression assumes the following form:

where S_GDP is the GDP of Singapore in USD at 2005 prices, OC is Chinese oil consumption mea-sured in 10000 tons of standard coal equivalent, OP represents the price of oil in USD per barrel, C_GDP measures the GDP of China in USD at 2005 prices, T_GDP is the average GDP per capita of Malaysia, Indonesia and the United States, also in USD at 2005 prices, and CRC represents the crude oil refining capacity of China in thousand barrels per calendar day units. Natural logarithms of each of the variables are used in the regressions to help more easily interpret the estimated coefficients. For example, if β1 in the above regression is estimated to be two, a 1 percent increase in Chinese oil con-sumption is estimated to lead to a 2 percent increase in Singapore’s GDP. The inclusion of control vari-ables in the regression enables a more accurate es-timation of the impact of each of the regressors on Singapore’s GDP, represented by the coefficients β1 through β5, by isolating their effects. For a trade-specific analysis, I then repeat the above regression using Singapore’s merchandise trade as the dependent variable:

where S_MT is the total value of Singapore’s mer-chandise trade measured in USD at 2005 prices, OC is Chinese oil consumption measured in 10000 tons of standard coal equivalent, OP is the price of oil in USD per barrel, C_GDP represents the GDP of China in USD at 2005 prices, T_GDP is the aver-age GDP per capita of Malaysia, Indonesia and the United States in USD at 2005 prices, and CRC is the crude oil refining capacity of China in thousand barrels per calendar day units.

Page 40: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

40 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

The OLS estimators from the above models are the best linear unbiased estimators, given that three key assumptions hold. First, the conditional distribution of the residual ut, given the dependent variables in the regression, must have a mean of zero, so that no omitted variable bias is present. The inclusion of control variables of oil prices, Chinese GDP and the GDP of Singapore’s major trade partners in the model helps to ensure the fulfillment of this require-ment. Secondly, all of the variables in the regression must be independently and identically distributed (i.i.d.). Finally, large outliers in the data should be unlikely, such that the kurtosis of all variables is fi-nite. This condition is fulfilled by the dataset being used. A stronger set of assumptions includes a fourth condition, that the residuals must be homoskedastic. However, the above regressions have all been con-ducted using heteroskedasticity-robust OLS regres-sions that do not require the last assumption. The second condition is the most problematic in this paper, as time-series data is being analyzed, and none of the variables fulfill the i.i.d. assumption. The key problem that may is that of spurious corre-lation, where variables in a time-series analysis may seem to be strongly correlated, but do not have any true underlying causal relationship. To avoid this problem and draw conclusive results, we must more closely investigate the dynamic causality between Chinese oil consumption and Singapore’s economic indicators. To undertake this analysis, I follow the meth-odology used in Zou and Chao’s study of the effects of Chinese oil consumption on China’s own eco-nomic growth (Zou & Chau, 2006). The paper uses the concept of Granger causality, which proposes studying potential causal effects existing between time series variables. If X is found to Granger-cause Y, it contains statistically significant information about future values of Y. Then, X can be considered a good predictor of Y, and X may have a causal ef-fect on Y. The Granger test equation takes the form of the following Autoregressive Distributed Lag (ADL) Model:

where β0 is the constant term, t indexes for time, p and q are lag lengths that are sufficient to ensure that ut is white noise, and ut is the residual with mean zero (Lütkepohl, 1993). The standard Wald F-test on all the coefficients from γ1 to γq is then used to detect Granger causality running from X to Y, using equation 4 given below.

where RSS1 is the sum of the squared residuals in Regression 3, RSS0 is the sum of squared residu-als in an autoregression for Yt that only includes lagged values of Yt as regressors, N is the number of observations, and F calculates the F-statistic for the F(N-2n-1) distribution. If this F-statistic is sta-tistically significant, we can conclude that X does indeed Granger-cause Y. However, as noted previously, the problem of spurious correlation may arise in conducting such a regression. In particular, if the time series contains one or more unit roots in its linear stochastic pro-cess, it is a non-stationary process and the results of any F-test based on this spurious regression diverge, leading to the rejection of almost any null hypoth-esis, given a sufficiently large sample size (Phillips, 1985). Thus, we must first test the stationarity of the three time-series variables of interest: Chinese oil consumption, Singapore’s GDP and Singapore’s merchandise trade. The augmented Dickey-Fuller (ADF) tech-nique can be used to detect the presence of one or more unit roots in a time series. If we can reject the null hypothesis that a time-series variable has a unit root, it is stationary. If we cannot reject the null hy-pothesis of a unit root on the variable itself, but can reject the null on its first difference, the time series is integrated of order 1, denoted by I(1). In general, the series is integrated of order d, denoted by I(d), if the series and all its differences up to the (d-1)st difference are non-stationary, but the dth difference is stationary. Upon determining the order of integration of each of the variables, we use the Toda-Yamamoto procedure for testing Granger-causality (Toda & Ya-mamoto, 1995). To do so, we first propose the fol-lowing ADL model:

where S_GDP is the GDP of Singapore in USD at 2005 prices, OC is Chinese oil consumption mea-sured in 10000 tons of standard coal equivalent and t indexes for time. Some of Singapore’s refined oil products are sold on the spot market, which consists mainly of one-off deals and thus would not be af-fected by lags. The majority, however, is sold under term contracts of varying durations, generally up to three years, so I would expect the number of lags in Chinese oil consumption that enter the regression to be a maximum of three (Platts, 2010). In the case of Singapore’s GDP, I follow the methodology used in Abeysinghe’s paper on forecasting Singapore’s economic indicators to propose an optimal lag of 3 years (Abeysinghe, 1998). The Bayes Information

Page 41: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 41

Volume xxiii. issue i.

Criterion (BIC) is used to confirm that the appro-priate number of lags for Chinese oil consumption (OC) is 3 and that for Singapore’s GDP (S_GDP) is also 3. Instead of running the Granger-causality test di-rectly on equation 5, the Toda-Yamamoto procedure corrects for the non-stationarity of the time series variables in the regression by including m additional lags of each of the variables in the regression, where m is the maximum order of integration among all the regressors, as established by the ADF tests con-ducted earlier. The transformed ADL model, with the extra lagged regressors included, takes the fol-lowing form:

where S_GDP is the GDP of Singapore in USD at 2005 prices, OC is Chinese oil consumption mea-sured in 10000 tons of standard coal equivalent, t indexes for time, and m is the maximum order of integration of the S_GDP and OC time series vari-ables. The inclusion of these additional lags reestab-lishes the asymptotic chi-squared distribution of the Wald F-statistic under the null hypothesis (Toda & Yamamoto, 1995). The key to the Toda-Yamamoto procedure is that the subsequent F-test used to check for Granger-causality only tests the coefficients on the regressors included in Regression 5, and not the m additional lags included in Regression 6. This F-statistic then enables us to test whether Chinese oil consumption Granger-causes an increase in Singa-pore’s GDP. The test for Granger-causality will then be re-peated to check whether Chinese oil consumption Granger-causes an increase in Singapore’s merchan-dise trade, using the same methodology outlined earlier, and equations 7 and 8 below:

where S_MT is the total value of Singapore’s mer-chandise trade measured in USD at 2005 prices, OC is Chinese oil consumption measured in 10000 tons of standard coal equivalent, t indexes for time, and n is the maximum order of integration among the two regressors. The lag estimate of 3 on Chinese oil con-sumption is based on the duration of the term con-tracts as previously described, while the number of lags on Singapore’s merchandise trade is based on forecasting models used by the International Mon-etary Fund (Parrado, 2004). These estimates were then confirmed by minimizing the Bayes Informa-

tion Criterion for each of the regressors.

V. Results and DiscussionThe OLS estimates obtained from Regression 1 and 2 are reported in Table 3 below. These regres-sions examined the correlation between an increase in Chinese oil consumption (OC) and Singapore’s GDP (S_GDP) and merchandise trade (S_MT) re-spectively, controlling for four other variables– oil prices (OP), Chinese GDP (C_GDP), the GDP of Singapore’s major trade partners (T_GDP), and Chinese refining capacity (CRC).

Table 3: OLS regression results for Regression 1 and Regression 2

The coefficient on China’s oil consumption (OC), regressed on Singapore’s GDP (S_GDP) in Regression 1, has a p value much lower than -0.05, and thus, is not statistically significant at the 5 per-cent significance level. There does not seem to be any statistically significant correlation, positive or negative, between China’s oil consumption and the GDP of Singapore. As discussed earlier, one poten-tial reason for this could be that an increase in Chi-nese oil consumption seems to be relatively unim-portant to Singapore’s economy as a whole since oil refining contributes to only 5 percent of Singapore’s GDP. In contrast, in Regression 2, the coefficient of Chinese oil consumption (OC) on Singapore’s mer-chandise trade (S_MT) carries a p value of 0.073. Although this is not below 0.05, it is significantly closer to the significance threshold of 0.05 than in Regression 1, and the use of a laxer 10 percent sig-nificance level would make the correlation between China’s oil consumption and Singapore’s trade a positive and statistically significant one. This find-ing fits in with the preliminary information pre-

Page 42: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

42 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

sented on the data in Figure 3, which shows how Singapore’s trade figures are more responsive than its GDP values to changes in Chinese oil consump-tion. In addition, since oil refining activities account for over 30 percent of Singapore’s total merchan-dise trade, but only 5 percent of Singapore’s GDP, it is indeed highly plausible that a correlation will be observable between Chinese oil consumption and Singapore’s trade, but will not translate into a statis-tically significant correlation with Singapore’s GDP as a whole. The other regressor of note is China’s own oil refining capacity (CRC). In both Regression 1 and 2, the p value on the coefficient on Chinese refining capacity is very large, so this variable does not seem to have a statistically significant correlation with the key indicators of Singapore’s economy analyzed in this paper. There could be several factors driving this finding. Firstly, Chinese refining capacity that is currently in operation could be used to serve dif-ferent markets from those that import Singapore’s refined oil products. With its strategic geographic location, Singapore is able to serve as a convenient exporting hub that can transport refined oil products via sea routes to coastal regions of Southern China, Malaysia and Indonesia. In contrast, China’s major new oil refining projects are located in far northern or interior provinces, such as Shandong, Xinjiang, or Gansu provinces (U.S. Energy Information Ad-ministration, 2014). These locations often neces-sitate less efficient over-land transportation routes to reach the regions that are currently serviced by Singapore’s refined oil products. Consequently, they may not pose significant competition to Singapore’s oil refining sector, offering a potential explanation as to why the observed expansion in Chinese refin-ing capacity does not correlate with a decrease in Singapore’s GDP or trade. Another key reason may be that, as an entrenched market player enjoying the first-mover’s advantage, Singapore already has large refineries that benefit from larger scales of production and greater energy efficiency in operations than many of the new re-finery plants in China and can thus produce refined oil products at a more affordable cost. In addition, while Chinese oil refining capacity is spread over various different provinces, Singapore’s oil refin-ing capacity is concentrated within a single hub, enabling new projects to leverage on existing facili-ties to reap technical economies of scale and gain access to the latest refining technology. In contrast, new Chinese refining projects struggle to meet pol-lution standards, severely impacting the viability of

refinery expansion. Last year, China’s two biggest oil producers, PetroChina and Sinopec, were banned from expanding their refining capacity due to envi-ronmental violations by their existing facilities (Ng, 2013). With Chinese leaders facing growing do-mestic and international pressure to curb industrial emissions, standards are expected to tighten fur-ther, adding more pressure to the profit margins of China’s oil refining sector. These limiting factors on the efficiency and profitability of producing China’s refined oil products may allow Singapore’s oil refin-ing industry to retain sufficient product quality and cost differentiation to continue to thrive in the face of expanding Chinese refining capacity. This could help explain why an increase in China’s refining ca-pacity fails to negatively impact Singapore’s GDP and trade. The results of Regression 1 and 2 thus seem to indi-cate that an increase in China’s oil consumption is correlated with an increase in Singapore’s merchan-dise trade at the 10 percent significance level, due to the positive coefficient on Chinese oil consumption, and the lack of a statistically significant impact on Singapore’s trade from Chinese refining capacity. However, this does not translate into any correlation with Singapore’s GDP as a whole, as the coefficients on both Chinese oil consumption and Chinese refin-ing capacity, when regressed on Singapore’s GDP, are not statistically significant. Figure 5 below sum-marizes the nature of the correlations observed from the regression results.

Figure 5: Flowchart representing empirically ob-served channels of effect between Chinese oil con-sumption, Singapore’s trade and Singapore’s GDP

The next component of the analysis focuses on in-vestigating potential Granger-causality of China’s oil consumption on Singapore’s GDP and mer-chandise trade. First, the Augmented Dickey Fuller (ADF) test statistics are calculated to determine the order of integration of each of the three vari-ables Chinese oil consumption (OC), Singapore’s GDP (S_GDP) and Singapore’s merchandise trade

Page 43: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 43

Volume xxiii. issue i.

(S_MT). Table 4 presents the ADF statistics calcu-lated for these three variables. The p value on the ADF statistics of all three variables are greater than 0.05, so we cannot reject the null hypothesis of non-stationarity for any of them. Since the p value for the first differences of both Singapore’s GDP (S_GDP) and trade (S_MT) are less than 0.05, using a 5 per-cent significance level, we can conclude that these two time-series variables are I(1), or integrated of the order 1. In the case of the variable for Chinese oil consumption (OC), it is only the p value of the second difference that is below 0.05, so it is I(2), integrated of the order 2.

Table 4: ADF Test Statistics for variables OC, S_GDP and S_MT

Having established that the maximum order of integration among the regressors is 2, we run Re-gressions 6 and 8 by setting the number of addi-tional lags, m and n, in each to 2. These regressions are reproduced below with this substitution for the reader’s convenience.

where S_GDP represents the GDP of Singapore in USD at 2005 prices, S_MT is the total value of Singapore’s merchandise trade measured in USD at 2005 prices, OC is Chinese oil consumption mea-sured in 10000 tons of standard coal equivalent, and t indexes for time.The results of these regressions are presented in Table 5 below. The bolded variables indicate the regressors that have been included in the F-test for Granger causality in each regression.

Table 5: OLS regression results for Regression 6 and Regression 8

Since none of the regressors in both Regression 6 and 8 have coefficients with a p value lower than 0.05, or even lower than 0.10, it seems that none of them are individually statistically significant at the 10 percent significance level. However, the test for Granger-causality calculates a Wald F-statistic to determine whether all the three lagged values of Chinese oil consumption (OC), excluding the addi-tional two lags that were included to correct for the non-stationarity of the regressors, are jointly signifi-cant, and this may be true even if none of the lagged values are individually significant. In Regression 6, the Wald F-statistic is 0.95, which is lower than the 5 percent critical value of 4.08. Hence, at the 5 percent significance level, we cannot reject the null hypothesis that the first three

Page 44: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

44 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

lagged values of China’s oil consumption (OC) are jointly insignificant and do not carry any predictive value for future values of Singapore’s GDP (S_GDP). Thus, we cannot conclude that an increase in Chinese oil consumption Granger-causes a change in Singapore’s GDP level. The F-statistic calculated for Regression 8, however, has an F-statistic of 15.66, which exceeds the critical threshold of 4.08. Therefore, at the 5 per-cent significance level, we can reject the null that the coefficients on the first three lagged values of China’s oil consumption (OC) are not statistically significant. These results indicate that an increase in China’s oil consumption Granger-causes a change in Singapore’s merchandise trade value in the short run, within a time-period of 3 years. The coeffi-cients on the three lagged values of China’s oil con-sumption in Regression 8 sum to 1.0991, a positive number. This agrees with our earlier finding that an increase in Chinese oil consumption is positively correlated with a rise in Singapore’s merchandise trade, but does not affect Singapore’s GDP. The implications of the results of this Granger-cau-sality test are noteworthy, but must be interpreted with caution. Firstly, it must be noted that the estab-lishment of Granger-causality indicates only that oil consumption in China has useful predictive value for future trends in Singapore’s trade, and does not necessarily prove a causal relationship between the two variables. There could be other, indirect chan-nels through which an increase in China’s oil con-sumption helps to boost Singapore’s trade, such as by leading to an increase in Chinese GDP, which in turn causes China to increase imports of Singa-porean products and Singapore’s merchandise trade rises. However, when this finding of Granger-causal-ity is evaluated in conjunction with the earlier OLS regression results, which demonstrated a positive correlation between Chinese oil consumption and Singapore’s trade even after controlling for other factors such as China’s GDP level, it does seem to suggest that Chinese oil consumption has a direct positive effect on Singapore’s total merchandise trade value. This is because of China’s increased demand for Singapore’s refined oil products, which are not easily substitutable by their Chinese-made counterparts due to issues of technical and environ-mental cost. Initially, this result may seem to validate the Singapore government’s emphasis on increasing investment in the oil refining sector, assuming its motivation in doing so is to boost trade figures.

However, beyond the sample period between 1990 and 2010 analyzed in this paper, the positive linear correlation between the two variables may not con-tinue to hold. It may be that Singapore’s oil refin-ing industry has already reached its efficient scale of production, in which case further expansion will only lead to diseconomies of scale and consequent-ly, rising costs and inefficiencies. Factors such as these must also be taken into consideration when making policy decisions on promoting investment of an industry. Another key issue to consider is whether the ongoing trend of China’s increasing oil consump-tion truly adds value to Singapore’s economy, or merely serves to inflate trade figures. The lack of a statistically significant correlation and Granger-causality between China’s oil consumption and Singapore’s GDP indicates that the positive effect of China’s growing demand for oil on Singapore’s trade may be diluted by the diversification of Sin-gapore’s economy such that its impact Singapore’s GDP is negligible. Due to the unique nature of Sin-gapore’s re-export driven oil refining, this industry is overrepresented in merchandise trade values as compared to GDP figures. Hence, even if the Sin-gapore government’s increasing investment into the oil refining sector leads to higher trade figures, if its ultimate motive is to increase the country’s real GDP level, it would be worthwhile to consider the limited direct impact that increasing Chinese oil consumption has on Singapore’s GDP.

IV. Conclusion

The empirical findings in this paper indicate that the net direct effect of China’s growing oil consump-tion on Singapore’s trade has thus far been a positive one. The results suggest that fears of the impact of growing Chinese refining capacity on Singapore’s refined oil industry may have been overstated, due to the different costs and technology of production employed by the two. As of yet, Singapore has re-tained a competitive advantage by virtue of its first-mover advantage, but it remains to be seen whether this will continue to hold in the future. Furthermore, while Singapore’s oil industry has a large impact on the country’s trade figures, it does not make a proportionate contribution to its GDP. Thus, the impact of increasing Chinese oil con-sumption on the Singapore economy as a whole, as measured by GDP, has been statistically insignifi-cant. This may be an important factor to consider in policy decisions to continue expansion of Singa-

Page 45: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 45

Volume xxiii. issue i.

pore’s oil refining sector. However, there may be other, more long-term benefits to further investment into Singapore’s oil refining industry that have not been captured by the variables considered in this study. For instance, expansion of the island’s oil refining activities may help to encourage the anchoring of oil trading and price discovery in Singapore too, due to the syner-gies that arise from the downstream integration of refining, logistics and trading activities within a sin-gle hub. The full impact of these currently develop-ing effects may only surface in the analysis of future data. In addition, the environmental advantages of continuing to develop Singapore as a regional oil-refining hub may not have been fully reflected in the analysis undertaken in this paper. As mentioned earlier, Singapore’s refineries enjoy larger scales of production and greater energy efficiency in op-erations than many of the new refinery plants being opened in China and other developing countries in the region. The concentration of oil refining in hubs such as Singapore, which have access to the latest pollution-minimizing refining technology, may thus also help to minimize the negative global environ-mental impact of these activities. Further research that investigates these ad-ditional effects and takes into account longer-term interactions between growing oil consumption in the region and Singapore’s economy may help to provide additional insights into the desirability of expanding Singapore’s oil refining industry in the face of rapidly evolving energy consumption trends. Although this paper has established that the recent growth in Chinese oil consumption has not led to a significant change in Singapore’s GDP, future de-velopments in domestic, regional and even global policy decisions on oil use and production may still make the expansion of Singapore’s oil refining in-dustry a socially and economically optimal venture.

V. References

Abeysinghe, T. (1998). Forecasting Singapore’s quarterly GDP with monthly external trade. Interna-tional Journal of Forecasting, 14(4), 505–513.

Abeysinghe, T. (2001). Estimation of direct and in-direct impact of oil price on growth. Economics Let-ters, 73(2), 147–153.

Directorate General for Trade. (2013). Trade in goods with Singapore. European Commission.

Retrieved from http://trade.ec.europa.eu/doclib/docs/2006/september/tradoc_113443.pdf

International Energy Agency. (2013). Oil Market Report. Retrieved from http://omrpublic.iea.org/currentissues/fullpub.pdf

Ken, W. L. (1960). The Trade of Singapore, 1819-69. Journal of the Malayan Branch of the Royal Asi-atic Society, 33(4 (192)), 4–315.

Leong, E. (2013). Energy Industry in Singapore: An Industry Background. Economic Development Board of Singapore. Retrieved from http://www.edb.gov.sg/content/edb/en/industries/industries/en-ergy.html

Lim, R. (2013). Singapore’s Merchandise Trade Performance, 2003-2012. International Enterprise Singapore, (March 2013), 8–13.

Lütkepohl, H. (1993). Introduction to Multiple Time Series Analysis. Berlin and New York: Springer-Verlag.

Ng, E. (2013, August 29). Oil giants barred from new projects after missing pollution targets. South China Morning Post. Retrieved from http://www.scmp.com/business/companies/article/1300306/chi-na-environment-ministry-suspends-some-approv-als-sinopec-cnpc

Parrado, E. (2004). Singapore’s Unique Monetary Policy: How Does It Work? International Monetary Fund Working Paper, WP/04/01.

Phillips, P. C. B. (1985). Understanding Spurious Regressions in Econometrics (Cowles Foundation Discussion Paper No. 757). Cowles Foundation for Research in Economics, Yale University. Retrieved from http://ideas.repec.org/p/cwl/cwldpp/757.html

Platts. (2010). The Structure of Global Oil Mar-kets (Industry Solution Papers). The Mc-Graw Hill Companies. Retrieved from http://www.platts.com/IM.Platts.Content/InsightAnalysis/IndustrySolu-tionPapers/oilmarkets.pdf

Rajan, R. S. (2009). Singapore: trade, investment and economic performance. Hackensack, NJ: World Scientific.

Straits Times. (2014, April 15). Singapore March ex-

Page 46: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

46 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

ports seen down amid China slowdown. The Straits Times. Retrieved from http://www.straitstimes.com/news/business/economy/story/singapore-march-ex-ports-seen-down-amid-china-slowdown-20140415

Toda, H. Y., & Yamamoto, T. (1995). Statistical in-ference in vector autoregressions with possibly inte-grated processes. Journal of Econometrics, 66(1–2), 225–250.

U.S. Energy Information Administration. (2013). Analysis Brief on Singapore’s Energy Markets. Re-trieved from http://www.eia.gov/countries/analysis-briefs/Singapore/singapore.pdf

U.S. Energy Information Administration. (2014). Analysis Brief on China’s Energy Markets. Re-trieved from http://www.eia.gov/countries/analysis-briefs/China/china.pdf

Yep, E. (2013). Singapore Seeks to Profit from Spe-cialty Chemicals. The Wall Street Journal. Retrieved from http://online.wsj.com/news/articles/SB10001424127887324874204578437940173430794

Zhang, W., & Yang, S. (2013). The influence of en-ergy consumption of China on its real GDP from aggregated and disaggregated viewpoints. Energy Policy, 57, 76–81.

Zou, G., & Chau, K. W. (2006). Short- and long-run effects between oil consumption and economic growth in China. Energy Policy, 34(18), 3644–3655.

Zweig, D., & Jianhai, B. (2005). China’s Global Hunt for Energy. Foreign Affairs, 84(5), 25–38.

Page 47: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 47

Volume xxiii. issue i.

I. Introduction

In 2001, a system of price fixing between auc-tion houses Christie’s and Sotheby’s came to an end due to government intervention. The story made headlines and cost each firm hundreds of millions of dollars in fines, but even after the end of the col-lusion the industry remains all but dominated by the two firms. We examine how these auction houses, which practice one of the most common and impor-tant forms of trading in human history, continue to control an industry that remains highly profitable and constantly growing. The modern auction market (as a subset of the larger market for art, which includes businesses that sell art directly) has the general appearance of a per-fect competition model. The market is large on both sides of the buying equation [Over 9200 fine art auc-tions were held in 2006, with only 842 being held by Christie’s or Sotheby’s (Artprice)]. Additionally, the auction services at all locations and levels of profit are essentially the same, including gathering buyers and sellers together and making transactions. The two main variables that change between firms and auctions are the quality of the art pieces being sold and the commission fees charged by the auctioneer. Lastly, because the art auction industry is rela-tively large and consistently active, the entry and exit barriers should be relatively easy to deal with. The labor force in this market tends to have rela-tively high liquidity, and high-quality art is easy to liquidate even outside the art auction market. There-fore, the puzzle that we examine is why the art auc-tion industry is essentially a duopoly in the face of so many incentives for other firms to enter the mar-ket.

Yaoyuan Liu, Haoran Chang, Tucker PhillipsUniversity of Wisconsin - Madison

Competition in the Art Auction Industry

EditorialThis is a fascinating exploratory piece on the forces at play in the high end art market. Having a piece auc-tioned at Sotheby’s or Christie’s is a rite of passage for treasured art. These two auction houses dominate the industry and give Yaoyuan Liu, Haoran Chang, and Tucker Phillips the opportunity to explore an under-discussed industry. Auctions, the Art Industry, and imperfect competition all become interesting economic topics to discuss. We hope you enjoy this informative and descriptive research on the art auction world.

Jacob Miller

We explore several potential reasons for this unique phenomenon, each of which deals with entry barriers inherent to the industry that may not be ob-vious at the beginning. Firstly, we find that the major firms utilize a price discrimination strategy in determining the sell-er’s commission that involves lowering the seller’s commission in the high-end market while keeping it relatively high in the low and medium-end markets. Secondly, we consider the reputations of the major firms as their own entry barriers. This idea deals with how consumers perceive art and how the ma-jor firms are uniquely situated to capitalize on these perceptions. As an extra facet of this argument, we look at how fine art is treated as an investment good, and how this plays off of the “Masterpiece effect” regarding high-end art. All of this is tied to the pres-ence of wealthy but unknowledgeable agents in the market due to the recent influx of wealthy foreign buyers. Thirdly, we discuss how economies of scale and economies of scope also relate to the interpre-tation of entry barriers in the art auction industry. Plus, how the effects of scale and scope are divided across buyers and sellers of art pieces (instead of the more traditional cost-sensitive consideration). Fur-thermore, we use the idea of “learning by doing” as an alternative tool to unravel the underlying puzzle. Because the art auction industry operates much differently than the usual manufacturing-based in-dustries one usually considers, it’s important to un-derstand how an auction actually works and who benefits in what way. The auction houses we are discussing commonly apply the “English Auction” method of bidding, wherein bidding begins low and bidders escalate their bids. When the bidding stops, the buyer who wins is required to pay the price set

Page 48: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

48 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

by his last bid (also known as the “hammer price”). The eventual cumulative price the seller needs to pay also includes the buyer’s premium and taxes (both of which are not included in the bidding price). However, this does not mean that the art pieces are necessarily sold at the hammer price, because the hammer price must be higher than the seller’s “re-serve price”. This price is considered confidential information and is known only by the auction house and the seller. Pieces that do not reach their reserve price in auction are thereafter referred to as “bought in” and are put up for later auction, sold elsewhere, or taken off the market. Apart from these public auctions, the firms also occasionally hold private auctions. These secondary auctions are somewhat significant: in 2012, private sales (which require a minimum consignment value of $50,000) accounted for approximately 10% of the total profit for Sotheby’s (Gioia). On the other side of the transaction, sellers need to pay for the commission (some percentage of the sale price that is paid to the auction company selling the piece). The seller’s commission varies, ranging from 5% to 15%. These seller’s commissions and buyer’s premiums are the main sources of profit for each auction company. For example, 91% of Sotheby’s profits come directly from these sources each year (Sotheby’s Reports). The market as a whole operates on the same principles, but non-leading firms are typically much smaller. There are a large number of auction houses worldwide, but most of them are locally focused and concentrate on less valuable art. Ultimately, there are only two dominant companies in this market: Christies and Sotheby’s. According to Artprices’ 2012 Market Trends report, Christie’s and Sothe-by’s accounted for about 71% turnover worldwide excluding the Chinese market. Christie’s is a private company, founded in 1766 by James Christie. By 2013, Christie’s had 53 offices in 32 countries and 12 salesrooms world-wide. Christie’s holds over 450 sales annually in more than 80 different categories of fine and deco-rative art. In comparison, Sotheby’s is a public com-pany, founded in 1744 in Britain. Its headquarters are now in New York and it has 90 branches in 40 countries. In 2013, Sotheby’s public and private sales accounted for 47% of the entire aggregate auction sales globally. There are also a few large or medium size art auction companies. As shown in the appendix 3, in 2012, Bonhams, Dorotheum and Getty Images follow the two dominant companies by a wide margin. The unique methods by which fine art is sold

in this industry forces auction companies to com-pete on the quality of the products rather than their price. In this case, the essence of competition in the art auction market is in trying to attract the clients who have the most valuable and unique art pieces for sale. Especially in the high-end categories, these pieces are exceedingly rare and, accordingly, ex-tremely valuable, which makes the competition over such pieces incredibly fierce. There are only a few elite collectors in the market, and they serve as the main source for Christie’s and Sotheby’s product lines. Charles Moffett, executive vice president and vice chairman of Sotheby’s worldwide impression-ist, modern and contemporary art department, said: “about 150 collectors worldwide can buy a painting for more than $20 million” (Esterow). In order to secure the sub-market in the high-end art category and keep the large clients, the art auction companies often forgo the seller’s commission fee or give them other benefits, such as sharing parts of the buyer’s premium. Typically, a seller has to pay about 10% commission on a $100,000 piece of art. However, if the art piece is worth more than $1 million or above, sellers don’t usually pay anything. For example, art-ist Jeff Koons’ “Balloon Dog (Orange)” was sold for $58.4 million in November 2013. The owner of the work, Peter M. Brant, said that Christie’s waived the seller’s commission and shared part of the buyer’s fees with him: “I was not required to give themany-thing from the buyer’s commission until it reached a certain price — which it did not make” (Bowley). Considering the costs, including purchasing in-surance to cover the costly investments and adver-tising each exclusive piece of art and its sale, Chris-tie’s ultimately suffers from waiving the seller’s commission. Similar commission waiving strategy can also be found in Sotheby’s operation. According to the Sotheby’s annual report in 2011 and 2012, it gave up more than $40 million annually in seller’s commissions. In a public letter from Daniel Loeb, the chief executive of the Third Point LLC, to Wil-liam Ruprecht, the chairman, president and chief executive officer of Sotheby’s, “Based on discussions with market par-

ticipants, it is our understanding that it has been Sotheby’s who has most aggressively competed on margin, often by rebating all of the seller’s commission and, in certain instances, much of the buyer’s premium to consignors of contested

works”. Because of these aggressive strategies in set-ting seller’s commissions on Christies and Sothe-by’s auctions in the high-end art market, high-end art sellers are compelled to sell their art pieces with

Page 49: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 49

Volume xxiii. issue i.

the two leading companies. Accordingly, new en-trants into the market are discouraged by the risk of burying such potential financial loss in the high-end market, thus creating an artificial entry barrier for new firms to try to tackle. As shown in the appen-dix 3, the highest prices for fine art were almost all the result of sales by the two dominating companies. Before 2008, Christie’s or Sotheby’s reliably sold the top 50 art pieces each year. On the other hand, auction companies can also charge much higher commissions on the low and medium-end segments of the art auction mar-ket. According to Christie’s CEO Steven Murhy, the sales of lower to mid-priced lots wind up con-tributing more profit than the high-end categories (Genocchio). The majority of the profits going to Christies and Sotheby’s come not from the high-end fine art that makes headlines but the medium and low-valuation fine art. As Murphy said, Christies’ 80 collecting departments handle goods priced as low as $200 and the category of art covering valu-ations up to $25,000 is rarely profitable against the operational costs to run it. In the high-end fine art category, the number of collectors is much smaller than the medium and low fine art markets, which allows these collectors to have more bargaining power in negotiating with the art auction companies. Facing this situation, Chris-ties and Sotheby’s discriminate the seller’s com-mission for different types of sellers. They set low commissions for the high-end sellers while forcing relatively high commissions on the medium and low-end sellers. We can see how this discrimination strategy works in the graph in appendix 1. As shown in the graph, the x-axis are the num-ber of sellers and the y-axis are the seller’s commis-sions. Since the high-end sellers have more bargain-ing power than the medium and low valuation art sellers, their expected seller’s commission is lower: the high-end seller’s expected commission is P1 and the medium and low seller’s expected commission is P2 (P1<P2). Suppose the art auction company set the single seller’s commission P0 (P1<P0<P2): they will lose their grasp of the high-end side of the market. If they set the single seller’s commission P0’ (P0’<P1), they will take these two markets and get the profit of area S1. If the auction company ap-plies a discrimination strategy that sets P0 for low and medium valuation fine art sellers and P0’ to the high-end sellers, they can take both these markets and receive a profit of area S2 > S1. Considering the limited number of collectors in the high-end category, companies can lower the commission for this type from P1 to Co, which is the

marginal cost of the company. Then the company can gain positive profit in the low and medium ends of the market but zero in the high-end. If Christies and Sotheby’s both apply this strategy, what happens if a new firm tries to enter into the market? Suppose a new entrant ventures into the market, but finds it impossible to enter into the high-end art market given that, facing the same cost as the incumbent, he will make zero profit. Therefore, the new entrant can only participate in the medium and low fine art auction markets profitably. If the new entrant wants to attempt to take over this market, the seller’s com-mission will ultimately dip lower than P0, which is the commission rate of the incumbent company. If P0 is larger than C0, the new entrant will have in-centive to enter into this market and make positive profits. Theoretically, Christies and Sotheby’s have to lower the seller’s commission to C0 for the low and medium sellers, which is the same situation as the strategy of setting the single commission price to C0. Despite this system, Christies and Sotheby’s do not set the low and medium markets’ fine art sell-ers to C0, yet they still earn positive profits in this market. Our claim is that they allow some new en-trants to enter into the low and medium market, but set high entry barriers in the high-end art market. However, the puzzle here is how they still can make positive profit in the medium and low-end market. Additionally, why do Christies and Sotheby’s intend to lose some profit in the seller’s commission for securing the high-end market? There are some alter-native explanations. Although Christies and Sotheby’s do not make profits on the seller’s commissions, they still can gain profit from the buyer’s commission. Appen-dices 5 and 6 show the Sotheby’s and Christies buyer’s commissions in 2013. That year, Christie’s charged up to 12% of commission for art pieces above $1.5 million, while Sotheby’s charged the same rate (12%) above the hammer price for art pieces selling for above $2 million. If they deviate from this aggressive strategy, they risk losing the entire market and settling for zero profit. By apply-ing the discrimination pricing though, Christies and Sotheby’s only lose the high-end seller’s commit-ment, but earn positive profit as a whole. Secondly, we contend that the high-end art mar-ket includes a spillover effect. The high-end market auction can benefit an auction company’s reputation and image in the media. The extremely expensive art pieces sold at auction can always attract people’s attention, which serves as its own kind of adver-tisement. The positive reputation can give auction

Page 50: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

50 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

houses more bargaining power in the low and me-dium seller’s commission, in which they can then earn positive profit. Although Christie’s and Sothe-by’s do not formally block entrance in the low and medium market, they have the advantage in these markets as a direct result of this reputational value. Also, considering the largeness of the low and me-dium markets, it is better to allow some companies to enter into the market rather than applying a com-plete deterrence strategy. This reputational aspect of the industry is hard to quantify, but it has a number of effects on how art is sold, collected, and marketed. On the buyer side of the market, perhaps the most obvious such char-acteristic is that art is a “Veblen” good: that is, the utility a buyer derives from any particular piece is directly proportional to the price that the buyer pays Thus, both utility and demand actually increase as the price does. The originator of this idea, economist Thorstein Veblen, saw that “there is…relatively lit-tle incentive to the exclusive possession and use of [art], except on the ground of [its] honorific charac-ter as items of conspicuous waste” (Veblen). Hence, utility from each purchase is derived not only from having the piece and viewing it, but also from the simple fact that the piece was costly to acquire. Additionally, art (especially high-end art) is characterized as being a credence good (that is, “consumers do not know which quality of a good or service they need” (Dulleck and Kerschbamer)). Dulleck and Kerschbamer examine such markets through a model where the price of both low and high quality goods (or, in our case, low and high “quality” art) does not depend on the actual qual-ity of the item. Since the experts who evaluate each individual piece possess more background informa-tion and market familiarity, they can set the expect-ed return for a piece such that p=c+h(C-c), where c is the actual value of the item if of a low quality, C is the actual value of the item if of a high quality, and h is the probability that the piece truly is of a high quality. Though an uninformed buyer or seller does not necessarily know the value of h, the art expert should be able to determine what category the piece falls into reliably. This incentivizes the art auction houses to overcharge for or overvalue each piece, and “consumers know this and expect the expert to charge the higher price” (D&K). The informa-tion gap between seller and consumer, based off the fact that neither buyer nor seller may truly know the value of the art piece, allows the art auction houses to serve as price-inflating middlemen. Therefore, the perceived value of art is largely based on price, and because Christie’s and Sotheby’s

are uniquely endowed with reputations for being the best in the market for retaining and utilizing art knowledge, both buyers and sellers are incentivized to deal primarily with the industry’s two main firms. Each one routinely shatters its own sales records on both individual and group lots. Christie’s alone “sold 58 works of art for over $10 million each dur-ing 2013 and achieved seven of the top 10 artwork prices in the auction market” (Christie’s), whereas Sotheby’s reported total sales of over $6 billion the same year (Sotheby’s Report). Together the two firms dominate this top-end of the fine art market, collectively filling the ranks of the most expensive single auctions every year. This means that both firms benefit to a dispro-portionate degree from the art market boom that has coincided almost exactly with the breakup of their price-fixing scheme. “There are a lot of new collec-tors, specifically international collectors, coming into the market” (Markley). Buyers from China, Ja-pan, and Europe have pushed the value of the art market up every year, and experts do not consider the recent surge a bubble. Rather, they see the surge as genuine explosive growth in direct contrast to the recovering US economy. That the market is growing so quickly at a time when it would usually shrink is evidence of long-term growth (Adam). Still, the distortionary aspects of all these new buyers play directly to Christie’s and Sotheby’s strengths. Buyers in the high-end segment of the fine art market “have enormous purchasing power.” (Rooney). These new customers are eager to buy, but they do not necessarily possess the same knowl-edge of art and its values as more seasoned bidders. This situation increases the information gap that al-lows prices to skyrocket. The high valuations benefit both the auctioneers (who increase both profits and reputations with each record-setting sale) and buy-ers (who derive more utility from art they perceive to be of greater quality). In particular, Christie’s re-cord-setting $7.13 billion in sales in 2013 were to a large extent due to a sharp spike in Chinese buyers: they made up over a fifth of the firm’s total sales (an increase of 63% from the previous year) (Chen). Im-portantly, both firms allow absentee bidding over the telephone, which gives foreign buyers opportunities to enter into even the priciest and most prestigious auctions easily and at no additional cost. Christie’s took advantage of these new markets even more di-rectly by opening an auction house in Shanghai in 2013, with an opening night that brought in $25 mil-lion in revenue (thereby shattering Christie’s own sales expectations) (Salesrooms & Offices). It is important to note that, though this trend

Page 51: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 51

Volume xxiii. issue i.

of growing prices increases the value of each piece of work, this does not necessarily have an affect on how art is treated once it is bought. Though art has an intrinsic aesthetic value, it is also used as an in-vestment good by both consumers and institutions (including banks and other financial institutions). This has become more and more true as investors look at the record prices set during recent art auc-tions and view them as signs of guaranteed, healthy returns (Adam). In general, art can be treated as an investment good that appreciates slightly in value over time (with an average annual return of 10%), but this only applies to art that routinely gets sold and resold. As tastes change, it’s possible that in-vestors may be left with “lemons” that were once “plums”. This only further incentivizes the market’s focus on the most expensive pieces, and discourages buyers from pursuing new talent in favor of the ac-cepted “norms” for high-end art: post-war and con-temporary artists (Melby). This high-end focus not only gives the two ma-jor firms a distinct market share advantage, but it also ties into the way reputation fuels concentration in the industry. There are multiple facets of this re-lationship, and each combines to posit the two firms as seemingly unshakable industry leaders. The two firms’ grasp on “masterpiece” items all but splits the high-end art industry into its own separate market, wholly removed from the “local art” products that small auctioneers rely on for a steady flow of in-come. Part of this has to do with location: New York is considered the de facto art center of the country, which gives auction houses in the city both a repu-tational and economic advantage. “Art not shown in New York loses value” because it cannot be ac-cessed by the “key critics, curators, dealers, and col-lectors” located in the city, and thus it loses out on the added value these people provide, as well as the prestige associated with the localized market there (Plattner). The information these people provide leads to more sales, as it creates a perception of definite value. A buyer of any type of art may be “totally at a loss to explain why one water-color painting is worth $300, another $3,000, and a third $30,000 while the size, shape, quality of materials and other visual attributes may not vary significantly” (Plattner). On the contrary, art auctions in New York are able to utilize the city’s combined artistic capital to give purchasers a sense of safety in their (often expensive) purchase that cannot truly be found any-where else in the country. This capital is, in part, the result of a first mover advantage on the part of the two firms. Unlike other

New York-based auction houses (including Doyle New York and Swann Galleries), both the Chris-tie’s and Sotheby’s auction houses are extensions of art businesses that have been major players in the world art market since the 18th century. This allows the two firms to “purchase assets at market prices below those that will prevail later in the evolution of the market” (Lieberman & Montgomery). Addi-tionally the high-end art market is inherently limited by the number of masterpiece-level works of art in the world. This translates into a direct advantage in preemption of scarce assets. It is directly tied to economies of both scale and scope. The two auction houses are extensions of worldwide art institutions that maintain a sales presence in all the major art centers of the world. Another way reputation devalues art outside a specified value range is that it disrupts the usual economic equilibrium. Pieces of masterpiece-level art are widely regarded as existing outside the usual bounds of the market. That is, they are expected to routinely outperform and give a larger return on their respective investments than lower quality art. A supportive idea is that “it’s always better to buy one $10,000 object than ten $1,000 objects” (Gins-burgh). This idea requires the assumption that mas-terpieces will retain their inflated value during eco-nomic downturns, whereas lesser art will not. Economies of scope also function within the in-dustry as an extension of the market’s reputational focus. Though both Christie’s and Sotheby’s are pri-marily art auction houses, they have vastly expand-ed the range of products they offer while still main-taining reputations as high-end businesses. Both firms have expanded into the jewelry, wine, and real estate markets, and these expansions help to accen-tuate the same positive features that have solidified their grasps of the art auction market. Their real es-tate listings range from properties valued in the hun-dreds of thousands of dollars all the way up to the hundreds of millions. Likewise, their wine listings span the range of directly purchasable bottles all the way up to single cases that have fetched $102,850 (Sotheby’s) and $476,280 (Christie’s), helping pro-pel, for example, Christie’s 2013 wine sales itself is $75 million alone. Additionally, both firms offer storage services that are advertised as extensions of the firms’ estab-lished roles as premium art businesses. Christie’s offers Fine Art Storage Services, staffed by “hand-picked experts” who “receive the same training as Christie’s team and understand the unique handling requirements of any property” (CFASS), whereas Sotheby’s offers only a single storage facility in

Page 52: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

52 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

Middlesex, England. Christie’s decision to spend $22 million on a New York-based storage facility (Cardwell) deters entry by serving as a gallery for the work it sells and by allowing customers to avoid the VAT fees associated with overseas art storage in Europe. As a result of all these reputational factors, the competition between the two firms (and others) is essentially a complex Nash equilibrium where each factor is highly correlated with all others. Since both Christie’s and Sotheby’s are some of the oldest firms in the auction industry, they are able to cash in on prominent first-move advantages (which are partial-ly dependent on and embodied in the reputation of each firm). Even after 300 years of industrial change and growth, first entry advantages appear to still play an important role as a function of reputation. Because Christie’s and Sotheby’s have remained in the industry for such a long period, it is natural for people to consider selecting their services first. Be-yond the long history, Christie’s and Sotheby’s also sell more in terms of both quantity and category. For example Christie’s has more than 80 sales catego-ries and Sotheby’s has over 60, including musical instruments and automobiles. These expansions into other markets mean that both firms benefit more from economies of scale and economies of scope. Firstly, it is better for an auction company to have a larger number of art pieces available for sale, since each expert and auctioneer has a certain level of working ability and willingness to provide labor. In order to maximize the utilization rate of workers, it is better to have more scheduled auc-tions per worker with more works for sale in each. Secondly, the fixed fee of holding each auction tends to be lower when auction houses have more auctions in the same time period or simply present more art pieces per auction. For example, renting a building in which to hold auctions for a certain time period is one of the major costs for each major firm, and these costs can be mitigated somewhat by selling more expensive works and works from dif-ferent categories. Thirdly, having more categories results in a lower cost of advertising as more items can be sold over fewer events. Once Christie’s and Sotheby’s start to reduce their costs relative to their competitors, there are more financial advantages for them in the long run. For example, these two leading companies can easily mimic the pricing strategies of smaller firms. Even adopt an aggressive pricing strategy, in order to compete with their opponents in the low and middle level markets while still earning profits overall. Moreover, because they usually do not have a commission fee for the high-end market

and at times even share part of the buyer’s commis-sion with the seller of the painting, individuals with a masterpiece will inherently prefer to work with the top established firms. In addition, the notable masterpieces that sell in the high-end market further increase the top firms’ reputations in the art auction industry. Even if it is very difficult to measure this sense of reputation and by how much it actually grows in real terms, it is cumulatively significant for both dominant firms. Besides this traditional viewpoint, we think the economies of scale and economies of scope devi-ate partially from simple analysis (i.e. analysis based only on the economies on the firm’s side of the market) and also interact with the consumer side of the art auction market. To understand this indus-try further, one must understand that the consumer side of the industry contains both buyers and sell-ers. This differentiates the industry from the usual system whereby the seller is some profit-focused, price-sensitive firm. For buyers of art pieces, the economies of scale effects of the auction houses are not very significant, but they have the potential to lower individual search costs for the each buyer. For example, a buyer with a general interest in a certain time period or style instead of a very particular tar-get will benefit from these economies of scale. If there are more paintings being sold in the auction the buyer will be more likely to both find a suitable piece and walk away with a piece of art despite the competitive nature of bidding. This cost lowering effects can also be found in economies of scope perspective. That said, when art auction companies have more departments from which to pull products, buyers, whose interests in specific pieces persist over relatively long time periods, will essentially have lower search costs. Therefore, they are more likely to choose the major firms with high levels of economies of scope. Based on the assumption of rational participants in the industry, both buyers and sellers of art pieces focus on utility maximization. This utility can be obtained through various actions outside of actually purchasing art pieces, such as simply browsing or attending auctions. Therefore, even for those buy-ers with specified targets, it can still be the case that they are more willing to choose those firms with higher level of economies of scope. One should realize that the analyses above are not valid for the high-end customers because they can get higher utility from other activities and their target art pieces are entirely unique. Regarding the seller side of the art auction, the economies of scale are more crucial than the economies of scope. In an

Page 53: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 53

Volume xxiii. issue i.

art auction, we should perceive buyers and sellers an integration, where both work in tandem to pro-duce a mutually beneficial outcome. Because of the effects of economies of scale and scope, auction houses will attract substantial amounts of buyers, which raises the probability of successful transac-tions. Therefore sellers will prefer large, established companies with such characteristics. A more critical argument for this is if there are more buyers, they will bid more aggressively so that the hammer price will converge to the actual buyers’ valuation accord-ing to the revenue equivalence theorem. Another ap-proach is that increases in either scope or scale will increase the reputation of each firm and sellers will consider such firms superior to others. Furthermore, since both Christie’s and Sothe-by’s have large effects of economies of scale and economies of scope and the learning by doing ef-fect in this traditional industry is mainly embodied in experts in the industry who will eventually retire or pass away, the knowledge carried by experts is localized on a person-by-person basis and does not necessarily spillover to the rest of the employees. As a result, these two leading firms will be con-stantly training new employees at all times in order to maintain their reputations as industry knowledge leaders. More importantly, since both firms have substantially larger effects of scale and scope than their opponents, they possess advantages in both depth and breadth of knowledge (or, at least, that is the reputation they present). Ultimately then, the duopoly in the art auction market seems to be a result not only of entry barriers established by the two major firms (including price discrimination and economies of scale and scope), but also barriers that are a result of the unique way in which art is purchased and perceived. These barriers make it all but impossible for firms to break away from local markets and into the national or global high-end art markets that Christie’s and Sotheby’s completely dominate. This helps explain why, given the exorbitantly pricy products both firms manage to sell on a regular basis, more firms have not entered into the market even after price fixing as a practice was excised from the equation.

References

Adam, Georgina. “Is art a good investment?” Cul-ture. BBC, Web. 5 May 2014. <http://www.bbc.com/culture/story/20131104-is-art-a-good-invest-ment>.

2006 Art Market Trends. Artprice, Print.

Bowley, Graham. “The (Auction) House Doesn’t Always Win.” New York Times. Web. 7 May 2014. <http://www.nytimes.com/2014/01/16/arts/design/christies-and-sothebys-woo-big-sellers-with-a-cut.html?_r=0>.

Cardwell, Diane. “A High-Tech Home for Multi-million-Dollar Works of Art.” NY Times. Web. 8 May 2014. <http://www.nytimes.com/2009/08/25/nyregion/25christies.html>.

Chen, George. “Chinese art buyers help Christie’s to record sales year.” China Business. South Chi-na Morning Post, Web. 5 May 2014. <http://www.scmp.com/business/china-business/article/1411102/chinese-art-buyers-help-christies-record-sales-year>.

Dulleck, Uwe, and Rudolf Kerschbamer. “On Doc-tors, Mechanics, and Computer Specialists: The Economics of Credence Goods.” Journal of Eco-nomic Literature 44.1 (2006): 5-42. Print.

Esterow, Milton. “Who Will Bid $50 Million?” ARTnews. Web. 7 May 2014. <http://www.artnews.com/2013/07/09/who-will-bid-50-million/>.

Genocchio, Benjamin. “Editor’s Point of View: What Do Christie’s and Sotheby’s Sacrifice to Win Top Consignments?” ARTnet. Web. 7 May 2014.

Ginsburgh, Victor. Handbook of the Economics of Art and Culture, Volume 1. N.p.: North Holland, 2006. Print.

Gioia, Lauren. 2012: An Exceptional Year for So-theby’s. N.p.: Sotheby’s, n.d. Print.

Kazakina, Katya. “Artist Murillo Denies Stardom With $50,000 Candy Crates.” Bloomberg. Business Week, n.d. Web. 5 May 2014. <http://www.bloom-berg.com/news/2014-04-30/artist-murillo-denies-stardom-with-50-000-candy-crates.html>.

Lieberman, Marvin, and David Montgomery. First-Mover Advantage. 1987. Print.

Loeb, Daniel. “Third Point Urges Sotheby’s Stock-holders to Support Its Slate of Independent and Highly Qualified Nominees.” Business Wire. Web. 7 May 2014. Melby, Caleb. “Study: You Should Invest 0% Of Your Assets In Art.” Forbes. Web. 5 May 2014. <http://

Page 54: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

54 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

www.forbes.com/sites/calebmelby/2013/11/19/ study-you-should-invest-0-of-your-assets-in-art/>.

Plattner, Stuart. “A Most Ingenious Paradox: The Market for Contemporary Fine Art.” American An-thropologist 100.2 (1998): 482-93. Print.

Press Release: For Immediate Release. : Christie’s, 22 Jan. 2014. Print.

Rooney, Ben. “Is there a bubble in the art market?” CNN Money. CNN, Web. 5 May 2014.<http://money.cnn.com/2013/11/15/investing/art-market-boom/>.

Salesroom & Offices. Christie’s, Web. 5 May 2014. <http://www.christies.com/locations/salerooms/shanghai/>.

Sotheby’s Annual Report 2011. Print.

Sotheby’s Reports 2012 First Quarter Results. Print.

Sotheby’s Reports Fourth Quarter and Full Year 2013 Financial.: Sotheby’s, 27 Feb. 2014. Print.

Veblen, Thorstein. The Theory of the Leisure Class.: Macmillan, 1899. Print.

Appendix

Appendix 1

Appendix 2

Appendix 3

Appendix 4

Page 55: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

Fall 2014 | The Visible Hand | 55

Volume xxiii. issue i.

Appendix 5

Appendix 6 (Sotheby’s Buyer’s Commission Rates)

Page 56: December 2014, Volume XXIII, Number I · 2019. 12. 8. · Sanjana Kanthimathinathan Nabeel Momin Manu Sharma Nivedita Kutty Vatsa Editors and Referees: Charles Anyamene Monica Cai

56 | The Visible Hand | Fall 2014

Volume xxiii. issue i.

rso.cornell.edu/ces