NBER WORKING PAPER SERIES EVIDENCE ON GROWTh,
NBER WORKING PAPER SERIESEVIDENCE ON GROWTh,INCREASING RETURNS AND THEEXTENT OF THE MARKETAlberto F. AdesEdward L. GlaeserWorking Paper No. 4714NATIONAL BUREAU OF ECONOMIC RESEARCH1050 Massachusetts AvenueCambridge, MA 02138April 1994We are grateful to Robert Barro, José Scheinkman. Andrei Shleifer and participants at theHarvard Growth Seminar for very helpful comments. Matt Botein and Melissa McSherryprovided excellent research assistance. We acknowledge financial support from the NationalScience Foundation and the Toner/Clark Fund. This paper is pait of NEER's researchprogram in Growth. Any opinions expressed are those of the authors and notthose of theNational Bureau of Economic Research.
NBER Working Papa #4714April 1994EVIDENCE ON GROWTh,INCREASING REI1JRNS AND THEEXTENT OF THE MARKETABSTRACTWe examine two sets of economies (19th century U.S. states and 20th century lessdeveloped countries) where growth rates axe positively correlated with initial levels ofdevelopment to document how these dynamic increasing returns operate. We find that openeconomies do not display a positive connection between initial levels and later growth; instead,closed economies do display this positive correlation (i.e. divergence). This evidence suggeststhat increasing returns operate by expanding the extent of the market (as in the big push theoriesof Murphy, Shleifer and Vishny (1989)). For U.S. states, we also find that larger marketsenhance growth by increasing the division of labor. Among LDCs, while more diversifiedproduction increases growth, diversification is negatively associated with openness for the pooresteconomies (as in the quality ladder theories of Boldrin and Scheinkman (1988). Young (1991)and Stokey (1991)). However, and despite the negative effect that openness has on the diversityof production and, thus, on growth, we find that openness still substantially increases growth forthese poorer economies.Alberto F. AdesDepartment of EconomicsHarvard UniversityCambridge, MA 02138Edward L. GlaeserDepartment of EconomicsHarvard UniversityCambridge, MA 02138and NBER
1. IntroductionFollowing Allyn Young (1928), much of the recent theoretical work on economic growth builds onincreasing returns to scale in production. Unlike models based on neoclassical production functions, thesemodels suggest that steady-state per capita growth rates are not independent of initial conditions.' Romer(1983 and 1986), Lucas (1988), Murphy. Shleifer and Vishny (1989), Rebelo (1991) and others useincreasing returns so that initial levels are positively correlated with later growth rates and the endogenousportion of growth continues indefinitely. Under particular assumptions, these models suggest that the theworld's lead economy will display divergence over time and often they also predict that the cross-sectionof the world's economies will also not show convergence.But empirical work on cross-country growth generally finds convergence, not divergence. BaumolMankiw, Rozner and Weil(1986), DeLong (1988), Barro (1991), Barro and Sala-i-Martin (1992) and(1992) document patterns of convergence in cross-country and U.S. data. Across countries, per capitaGDI's do not converge unconditionally between 1960 and 1985, but they do converge once we conditionon variables (such as education) that determine the steady state level of income per capita. This evidencewould seem to contradict modern endogenous growth theory.'In fact, the well-documented conditional convergence of Barro and Sala-I-Martin (1992) and others ismore relevant for tests based on the neoclassical growth model. Endogenous growth models predictunconditional, not conditional, divergence.' Most of these theories also predict that increasing returnsshould operate only during specific time periods (i.e. during industrialization) or 'in specific countries.We use the term neoclassical to refer to production functions that display diminishing returns toscale.'Growth theorists (see, for example, Romer (1986)) tend to avoid this contradiction by emphasizingthat divergence need only appear in the time series evidence of the lead economy. In many modelsdivergence is not predicted in a cross section.'Technically, the issue is whether or not savings or schooling behavior should be treated as anexogenous variable (as Solow (1957) treats Investment), or whether they should be considered asendogenously determined by other initial conditions. Barro (1991) takes a partial approach, treatingschooling as only a right hand side variable but treating investment as endogenouslydetermined by otherinitial conditions.
Instead of testing implications of the neoclassical growth model against the alternative hypothesis ofendogenous growth, we proceed by testing the hypotheses of endogenous growth models against eachother and against the alternative hypothesis that increasing returns never operate.Our primary question is whether increasing returns occur because higher Initial GD)' acts to increasethe demand for national products. Recent papers such as Becker and Murphy (1993) or Murphy, Shleiferand Vishny (1989) predict that growth %Ilows initial wealth because that wealth creates a market forcertain activities. According to these models, higher initial income should increase later growth ratesmore strongly when an economy is closed than when an economy isopen. In open economies, aggregatedemandis fixed by world markets and higher initial levels of income do not change effective demand.In closed economies, national income determines demand.We address this question with two samples of economies that display wacondllionaldynamic increasingreturns (U.S. states in the 19th century and the 65 poorest countries in 1960).' In our country samplewe use income and urbanization to measure growth and the share of trade in GD)' to measure openness.Across U.S. states, we use the state's levels of urbanization (following De Long and Shleifer (1993)) andmanufacturing to measure development. Nineteenth century income data for U.S. states is neither reliable(despite heroic efforts by Easterlin and others) nor theoretically appropriate in economies withconsiderable amounts of migration between states. For the states, we use physical distance to majorregional ports (and, thus, to major east coast markets)' and regional railroad development as our basicmeasures of openness.'Across all of our samples, we find that increasing returns axe important for closed no:open economies.This evidence suggests that increasing returns operate by expanding the extent of the market as in big'The importance of these samples immediately tells us about that there are places where increasingreturns do operate.'Such physical measures could also be used for 20th century data for U.S. states or cross-countrycomparisons. However, the vastly improved nature of transportation makes reliance on geography muchless palatable in the age of the airplane, truck and high speed ocean transport. See Pred (1980) for amore detailed discussion of these issues.'Regional rail has the useful attribute of giving us time series as well as cross-sectional variation.2
push theories (Rosenstein-Rodan (1943) and Murphy, Shlelfer and Vishny (1989)) or theories wheregrowth derives from the division of labor (Becker and Murphy (1993)). This finding does not supportthe implication of quaiity ladder arguments that openness is particularly damaging for economies at thevery bottom of the income distribution (see Stokey (1991) or Young (199!)).We explore ftzrther the relationship between initial levels, openness and growTh with a variety ofdecompositions. For both the nineteenth century U.S. data and the world data, the division of labor spursgrowth. In the U.S. data, initial GD? and openness increase growth in part by increasing the divisionof labor. For world data, initial GDP raises the division of labor, but openness deters it (as in Stokey(1991)). However, while openness acts against growth in poorer countries by reducing the division oflabor, the overall connection between openness and growth for the poorest nations is strongly positive - the direct positive effects of openness overwhelm the indirect effect that works through lost diversity.Section Il discusses the relevant theories. Section III describes the data. Section IV presents our basicempirical results starting with the openness/increasing returns connection. The next two sections presentour decompositions. Section VII concludes.II. Theories of Increasing ReturnsThis section presents the relevant implications of four major sets of theories of endogenous growth.lnformatiortoi AccessIn Romer (1983, 1986), endogenous growth is obtained with an aggregate production functioo thatexhibits increasing returns that are eiternal to the investing finn. Knowledge-based growth theories maypredict that openness spurs growth because of the connection of trade to information transmission (e.g.reverse engineering of traded goods, the Italian merchant Marco Polo, the Portuguese traders in 16th3
century Japan, or the innovations brought by European traders to native Americans).' But even ifopenness increasesaccess to a wider set of ideas for all countries,it is not obvious bow it interacts withgrowth at different levels of development. Two effects seem likely to drive the cross-effect betweencould make it easier to take advantage of new ideasopenness and initial GD?: (1) bigher human capitalso more developed areas may benefit most from the exposure created by trade, and (2) the new ideasbrought by trade might be less new to more developed countries. Evidence on the openness-initial GDPcross-effect can help determine the relative importance of these two effects but it cannot be used as a testfor the roleof informational spillovers more broadly.Big PushesA second theory of increasing returns is the big push theory of Rosenstein-Rodan (1943) and Murphy,Shleifer and Vishny (1989). This theory emphasizes the importance of coordination and demandspillovers in generating increasing returns and multiple equilibria. Technically, the Murphy a a!. modelsalso rely on the existence of fixed costs (and hence increasing returns) in technology, but the emphasisof their work lies on the role that increasing levels of income play in creating a larger market forindustrialized output.According to the theory of the big push, within closed economies more initial wealth creates moredemand for industrialized products which in turn induces firms to pay the fixed costs of growth. Bycontrast, in small open economies the demand curves facing local producers are set by world markets anduninfluenced by local wealth.' The big push theory predicts that growth will be related to initial wealthand access to world markets but also that the interaction between wealth and openness will be negative'Of course, some European innovations, such as the Inquisition, may not have been good for growth.'This type of effect might be part of the explanation for the smooth post-war Japanese business cycle.This emphasis on pecuniary externalities links growth with Keynesian (1936) macroeconomics. A naturaltest of the importance of pecuniary externalities for business cycles is to compare output volatility(relative to world output) for open and closed economies.4
as openness eliminates the linkage between initial wealth and effective demand. We also test for theimportance of the big push by checking if the extent of the market works by increasing the rate of growthof physical capital (where fixed costs are most likely to matter).Diversity and SpecializaüonA third set of increasing returns theories with implications for openness emphasize the role ofspecialization.The connection between the division of labor and economic progress is typicallyassociated with Smith (1976). Becker and Murphy (1993) argue for the role that the division of laborplays in increasing both the level and the growth rate of income overtime.' A finer division of laborcan speed up growth because concentration in a single task might facilitate innovation and learning bydoing (as Smith suggested). The costs of acquiring new skills might also lower as the range of tasksinvolved diminishes.'0Smith's famous dictum is that the division of labor is limited by the extent of the market. If that istrue, growth should be connected to initial levels only in closed economies. In open economies, theworld market is what determines the extent of the market. These models predict the same negativeinteraction between openness and initial wealth as the big push theories, but they also predict that thepositive effect of the extent of the market on later growth should lessen when we control for the divisionof labor.'Becker and Murphy emphasize the importance of coordination as opposed to actual market size forthe division of labor. However, their model also allows for the standard Smithian effects of market size.These effects become particularly close in spirit to their work when the statistical returns to scale of largermarkets are taken into account, i.e. when larger markets work by diversifying Idiosyncratic demandshocks to particular consumers.' A supposed advantage of assembly lines is the ease of training assembly line employees.S
The Qualisy Ladder: SpecialWng in the Wrong ProductsA final set of theories that offer predictions on the relationship between increasing returns and opennessare the quality ladder arguments of Boldrin and Scbeinbnan (1988), Stokey (1991), Grossman andHeipman (1991) and Young (1991), and the 19th century protectionists List (1856) and Rae (1834).Quaiity ladder theories are tied to the division of labor theories in that they argue that the range of goodsproduced is critical to growth, but in the quality ladder theories openness lessens diversity in production.Divisionof labor theories emphasize the importanceoftbehigberdemand forabroader rangeof productsthat comes about with increased openness; quality ladder models emphasize bow openness raises theforeign supply of these products, and thus lessens the effective demand for the domestic production ofthose products.Under quality ladder arguments, openness is particularly damaging for the extremely poor countries.With free trade, less developed countries will tend to specialize in low growth activities that are intensivein the use of natural resources or unskilled labor, thereby allowing middle income countries to reallocatetheir resources away from these low growth activities)' In Stokey's words "if the industries in whichthe less developed country has a static comparative advantage are industries in which there are limitedopportunities for learning, then the effect of free trade is to speed up learning in the more developedcountry and to slow it down in the less developed one." Therefore, and contrary to big push and divisionof labor theories, quality ladder arguments predict that initial income will be more closely associated withlater growth for open economies than for closed economies.III.- The DataThis section describes our data sets and their construction."High income countries may in fact also reduce growth as they reduce their production of highgrowth products and cater more to world tastes.6
Construction oft/ic Data Setsour cross-counuy data set was constructed using several different sources. The data on urbanizationwere assembled by hand from bard copy, and come from the 1988 edition of the Frospeas of Worldllrbwdzationi' The cross-countzy data on population and real per capita GD? are from the Barro andWolf (1991) data set. The trade data are from theWorld Bank's World Tables, and consists on importsand exports of goods and non-factor services. Data on educational attainment is from Barro and Lee(1993). The data on road infrastructure is from Canning and Fay (1992), and the land area data comefrom the 1986 edition of the FAQ Production YearbookFor the U.S. states, we also used a variety of different sources. The data on state population.urbanization and labor force are from the Historical Statistics oft/ic United States (1976). For populationin the state's main city, we used the 1980 Census and several issues of the Statistical Abstract oft/icUnited Stases. For some states, we used direct 19th century census data. The data on the labor forceengaged in manufacturing in 1880 and 1890 is from several issues of Statistical Abstract of the UnitedStates. For earlier years, we used the 1840 to 1870 censuses."The railroad data for 1860 to 1890, and the data on distance from the state's main city to the mainregional city are from the Ssatisticoi Abstract oft/ic United Stases. For each port, the relevant regionalport was either New York, San Francisco or New Orleans, whichever was closer. The railroad data for1840 and 1850 is from Wicker (1960). Literacy data are taken from the U.S. censuses. We had no12Data are available only for countries or areas with two million or more inhabitants in 1985."A problem with the U.S. census labor force and manufacturing data is that the populatioà covereddid not remain invariant during our sample period. Thus, while the 1840 and 1870 censuses covered thewhole population, the 1850 census covered the free male labor force above 15 years of age only, and the1860 census included free females and extended the age limit to 10 years or older. We dealt with thisproblem by obtaining census estimates of the slave population. To. construct labor shares inmanufacturing, we assumed that all slaves of 15 years of age and older were In the labor force, and that15 percent of them were in manufacturing (we based this figure on Sokoloff (1982)). Before thesecorrections, Southern states displayed wild variations in their manufacturing shares. We also tried alteringour assumptions about slave labor force participation rates and shares In nianufactur'mg slightly but noneof our results seemed sensitive to these alterations.7
choice but using data on white literacy only as before 1860 the census provides no Information on literacyrates for the slave population. Finally, we gathered data on over 300 hundred occupations from the 1850and 1810 censuses.Our U.S. data thus covers the deódes 1840-1890. Data was not collected before 1840 becauseofavailability problems. We stopped in 1890 because (1) massive immigration to eastern cities potentiallybiases our results, (2) rail development had become extremely comprehensive by 1890 so variation acrossregions became less meaningfol, and (3) by 1890 the eastern states had achieved a similar level ofdevelopment to the most developed nations incross-countryoursample. Moreover the period 1840-1890is typically considered the era of America's big push?Descripilon of the DataTables Ia and lb show the five fastest and five slowest growers in our cross-country sample ftioth intenns of GD? and urbanization) and the corresponding initial levels of the relevant variable. While theaverage initial income of the fastest growers is about 150 (in 1980 dollars) higher than that of theslowest growers, the group shows considerable heterogeneity. It includes both relatively well-off countriesas Malta and extremely poor ones as Lesotho. This is not the case with the slowest growers; all of themare in Sub-Saharan Africa.In terms of urbanization, the distinctions are more clear-cut. The average level of initial urbanizationfor the fastest growers is about double that of the slowest ones. Korea is the fastest grower on bothcounts,' and only five of the twenty countries are outside Sub-Sabaran Africa. Table Ic shows the fivemost and five least urbanized U.S. states in 1840, and table Id does the same thing for 1890. Table leshows the largest and smallest spurts in urbanization growth. Table If shows the largest and smallestchanges in urbanization over the 1840-1890 sample." There is a strong correlation between urbanization and Income growth across countries. Webelieve this fact supports our use of urbanization in the U.S. regressions.8
IV. Evidence on Increasing Returns and the Extent of the MarketThere are two major ways in which our estimation differs from more standard forms, e.g. Barro(1991): (1) we often use urbanization not income as our measure of development, (2) we focus onunconditional not conditional regressions.Urbanization vs. IncomeOur emphasis on urbanization (and manufacturing) over income for U.S. state data goes against theprevailing methodology and, admittedly, urbanization is often a poor proxy for economic development."However, the standard data source on state income levels (Easterlin (1960 used by Barro and Sala-iMartin (1992)) is available oniy at 40 year intervals, has measurement problems, and is in nominal dollar;(so differences across states might not reflect local price level differences).On the contrary,urbanization is (1) available every 10 years, (2) a simple, reliable measure, (3) invariant with respect tolocal price indices, and (4) reliably connected with economic development (see Bairoch (1988)).We also favor urbanization over income in the 19th century U.S. because intrastate mobility shouldeliminate any welfare differences across states. The income differences that do exist should represent acombination of unobserved heterogeneity and compensating differentials. The high incomes earned in19th century western states are much better interpreted as a compensation for the danger of the frontierand tediousness of life away from the eastern seaboard than as an index of economic development. Tablela shows the five least urbanized states of the U.S. in 1840. Without exception these states representsome of the least developed areas of the United Sta;es in this period.U The exact model that we have in mind is spelled out in the estimating framework section of thepaper.9
Condirionoi vs. (Jnconditionoi C.onwrgenceWe first focus on unconditional convergence rather than on the more traditional examination ofconditional convergence. This focus is appropriate since the four theories described above concernunconditional increasing returns. An advantage of the unconditional regressions Is that they are less proneto measurement error blase. When regressing GDP growth on initial GDP, measurement error createsboth the standard bias, which lowers the absolute value of the coefficient on initial GDP,1' but whenmeasurement error is i.i.d., it also lowers the coefficient on initial GDP byVar(Measuremen, Error)Var(JnifloJ GD?)(I)When conditional regressions are run, this second bias becomesVar(Measzjremem Error)Vor(lnifial GD? onhQgonollzed with respect to the other controls)(2)Since the denominator in (6) might be substantially smaller than that in (5), the bias towards convergencemight be much higher in conditional regressions. In the case of standard growth regressions, the biastowards convergence more than triples when going from unconditional to conditional regressions."Estinzo.sing FrameworkThe appropriate model for using urbanization as a proxy for development is one with two sectors: aprimary, unurbanized, agricultural, or low technology sector, and a secondary, urbanized, ormanufacturing one. Aggregate production in each state is given byThis coefficient is given byCov(GDP aonge,Ininai GD!')Var(InizjoJ GD!')The first bias works by raising the denominator. The second bias operates by lowering the numerator."This extra bias may explain why divergence appears in unconditional regressions only.10
A, (L; L41'(3)where A, is the overall level of productivity in the state, L,, and L2 are the quantities of labor in theprimary and secondary sectors, a measures the importance of the developed sector (the degree ofdevelopment), and s, represents some sort of state specific congestion. In equilibrium, the marginalproduct of labor will be equalized across states and sectors, which implies that the share of total outputin the secondary sector will be given bya-(4)Since we are primarily interested with changes in the structure of the economy, i.e. it's developmentfrom agricultural to urban, we need only lool at changes in the shares of population in each sector andinterpret them as changes in the coefficients of the Cobb-Douglas. More specifically, we look ata41.1ajj —f(a,1,O,)(5)where C),, represents the openness of economy I at time 1. We are particularly interested in the crosseffect between a,, and 0,. The specific functional form that we run isa fi0 1 cr4,. 2 O, a4, O, e,,,andwe are mostly interested with the sign of ,(6)the cross-effect between openness and initialdevelopment. All the regressions are weighted by initial population.World DataRegression (I) in Table 3 shows the raw divergence relationship for a cross-section of countries between1960 and 1985. The countries included in our sample are all those countrieswith Incomes of 1980 USS1,500 or less in 1960. There is a total of 65 countries in this basic sample. The relationship between11
initial levels and subsequent growth rates is fairly well known." Figure 1 shows this basic relationshipvisually. The coefficient of 0.019 indicates that an increase of USS 100.00 in 1960 increases the averagegrowth rate by 0.19 percent per year.Regression (2) shows our openness measure and the cross-effect between openness and growth. Asdiscussed earlier, this openness measure is flawed by Its endogeneity, I.e., trade is not exogenouslydetermined. However, the results show our basic point in a powerilil way. The pure effect of opennesson growth is moderatelypositive. A one standard deviation increase in the share of total trade to GD?increases the growth ratepercentage points(at the average initial level of GD? per capita in 1960 of USS 740) by 0.34per year (0.2 standard deviations).The cross-effect between GDP and openness is very strong. For an open economy with a share oftrade in GD? of 0.49 (slightly above the mean), there is no relationship between GDP and GD? growth.For a low trade, closed economy (with a trade share of 0.22, one standard deviation below the mean),a USS 100.00 increase in the level of initial GD? raises the growth rate by about 0.19 percentage pointsper year.Figures 2 and 3 show this result visually. Figure 2 shows the relationship between growth and initialGDP in low trade closed economies. Figure 3 shows that such relationship does not hold for hightradeopen countries. In regression (3), we introduce continent dummies and we control for primary schoolenrollment in 1960. The magnitude of the cross-effect rises once these controls are included." For asmaller set of countries, regression (4) reproduces regression (3) but using the initial share of trade inGD? instead of the average over 1960 to 1985. The results are consistent with those of previousregressions.Regressions (S)-(7) perform the same exercise as regressions (1X3) but using urbanization rates as ameasure of development. Regression (5) shows the raw divergence relationship for the same cross-sectionof countries between 1960 and 1985. The coefficient on initial urbanization indicates that, in oursample,"It can be seen in Figure 2 of Barro and Sala-I-Martin (1992)."We have also run these regressions with non-linear specifications of GD? as an explanatoryvariable and the results remained almost unchanged.12
a :10 percent increase in the initial level of this variable is associated with 4 percentage points fasterincrease in urbanization over the period.Regression (6) shows that once again the cross-effect is negative and strong. For, a moderately openeconomy with a share of trade in (DP of 0.46 (3 percentage points above the mean), there is norelationship between initial urbanization and subsequent changes. For a relatively closed economy (witha trade share of 0.22, one standard deviation below the mean), a 10 percent increase in the level of initialurbanization leads to a 4 percent increase in urbanization over the period.Figure 4 displays this strong positive relationship for the closed economies in our 65 least developedcountries. Figure 5 shows that there is no relationship between changes and initial levels for the openeconomies. Again, regression (7) verifies that our results are robust to controlling for regional andeducational variables. Regression (8) shows that our results are not sensitive to using the share of tradefrom 1960 to 1985 as our measure of openness.U.S.DataTable 4 contains similar regressions for our U.S. states sample. This table shows results for a pooledsample of states over the period 1840-1890. The decade 1860-1870 has been eliminated due to the CivilWar?' We have included fixed effects for each decade and allowed for correlation across decades inthe shocks to states by estimating a stacked set of growth regressions with SUR techniques (state specificrandom effects are a particular form of this methodology with an assumed form of correlation acrossdecades). The SUR methodology ensures that a state's growth rate between 1870 and 1880 aid a state'sgrowth rate between 1880 and 1890 are not treated as independent observations.'Our dependent variable is the decadal change in the share of urbanized population in the state. Thefirst regression in Table 4 documents the basic positive relationship between urbanization growth andinitial levels of urbanization. The time dummies tell us that the 1840-1850 and the 1880-1890 decadesr The results become substantially stronger if that decade is included.31 In fact there is not that mucb correlation between decadal growth rates across states.13
were the periods of strongest urbanization growth of the second half of the 19th century. The coefficienton initial urbanization in regression (9) Is positive and highly significant, and indicates thata 10 percentincrease in the amount of initial urbanization leads to about I percent increase in the share of urbanizedpopulation over a ten year period.Our first measure of openness is a distance dummy which takes a value of 0 if a state wa
growth. In the U.S. data, initial GD? and openness increase growth in part by increasing the division of labor. For world data, initial GDP raises the division of labor, but openness deters it (as in Stokey (1991)). However, whileopenness acts against growth in poorer countries by reducing the division of
the authors and do not necessarily reflect the views of the National Bureau of Economic Research. NBER working papers are circulated for discussion and comment purposes. They have not been peer-reviewed or been subject to the review by the NBER Board of Directors that accompanies official NBER publications.
NBER WORKING PAPER SERIES UP FROM SLAVERY? AFRICAN AMERICAN INTERGENERATIONAL ECONOMIC MOBILITY SINCE 1880 William J. Collins Marianne H. Wanamaker . are those of the authors and do not necessarily reflect the views of the National Bureau of Economic Research. NBER working papers are circulated for discussion and comment purposes. They have .
NBER WORKING PAPER SERIES UNIONS AND INEQUALITY OVER THE TWENTIETH CENTURY: NEW EVIDENCE FROM SURVEY DATA . Elena Marchetti-Bowick, Amitis Oskoui, Paola Gabriela Villa Paro, Ahna Pearson, Shreya Tandon, and Maryam Rostoum. We have benefited from comments by seminar participants at . groups
Tax Evasion at the Top of the Income Distribution: Theory and Evidence John Guyton, Patrick Langetieg, Daniel Reck, Max Risch, and Gabriel Zucman NBER Working Paper No. 28542 March 2021 JEL No. D31,H2
NBER WORKING PAPER SERIES WORKER OVERCONFIDENCE: . paper focusing on overconfidence. The other paper (Hoffman and Burks, 2017) studies the impact . the authors and do not necessarily reflect the views of any of the research funders or the National Bureau of Economic Research.
5 East Asia Seminar in Economics 17 (NBER and others) June 2006, Hawaii. TRIO Conference on International Finance (NBER, CEPR and TCER) December 2005, Tokyo. NBER Summer Institute, International Finance and Macroeconomics July 2005, Cambridge. East Asia Seminar in Economics 16 (NBER and others) June 2005, Manila. A
SMB_Dual Port, SMB_Cable assembly, Waterproof Cap RF Connector 1.6/5.6 Series,1.0/2.3 Series, 7/16 Series SMA Series, SMB Series, SMC Series, BT43 Series FME Series, MCX Series, MMCX Series, N Series TNC Series, UHF Series, MINI UHF Series SSMB Series, F Series, SMP Series, Reverse Polarity
3022 Broadway, 623 Uris Hall New York, NY 10027 and NBER apb2@columbia.edu Maya Rossin-Slater Department of Health Policy Stanford University School of Medicine 615 Crothers Way Encina Commons, MC 6019 Stanford, CA 94305-6006 and NBER mrossin@stanford.edu Christopher J. Ruhm Frank Batten Scho
