Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
This PDF is a selection from a published volume from the National Bureau of Economic Research Volume Title: NBER International Seminar on Macroeconomics 2007 Volume Author/Editor: Richard Clarida and Francesco Giavazzi, organizers Volume Publisher: University of Chicago Press ISSN: 1932-8796 Volume URL: http://www.nber.org/books/clar07-1 Conference Date: June 15-16, 2007 Publication Date: January 2009 Chapter Title: Comment on "International Portfolios with Supply, Demand and Redistributive Shocks" Chapter Author: Refet S. Gürkaynak Chapter URL: http://www.nber.org/chapters/c3011 Chapter pages in book: (277 - 281) Comment Refet S. Gurkaynak,BilkentUniversityandCEPR Introduction Researchon internationalportfolioholdings and flows have been reenergized with the introductionof data on gross portfolio holdings and analyticalmethods to solve for financialholdings in models with incompletemarkets.This paper adds to that literatureby addressingtwo of the most interestingstylized facts, described in the following, from within an innovativemodeling framework.It is an importantpaperthat asks the right questions and provides an initial answer using novel model solution methods. My discussion will first point out the modeling devices that help the authorsexplainthe stylized factsand then presenta briefnote on the calculationof the labor share,the varianceof which is used in the calibration of the model. Stylized Factsand the Model The paper,beginning with the abstract,mentions three stylized facts of industrialized economies to be explained: (a) portfolio holdings are biased toward local equity; (b) internationalportfolios are long in foreign currencyassets and short in domestic currency;and (c) the depreciationof a country'sexchangerateis associatedwith net externalcapital gains. There are actually two stylized facts to be explained as (c) is directlyimpliedby (b).If a countryis long in foreigncurrencyassets and short in domestic ones, a depreciationof its own currencywill lead to capital gains on those holdings. Indeed, the authors only focus on obtaining(b)as a model predictionand treat(c)as being obviously implied by this. Themodel is presentedin two steps. First,thereis a completemarkets 278 Gurkaynak version with two shocks and two assets that builds intuition into the workings of the mechanisms.Then a third shock is introduced,leading to marketincompleteness,in which case the model is solved using the Devereux and Sutherland(2006)method. The qualitativepredictionsof the model are developed in the complete marketscase, on which I will elaborate. In the completemarketscase thereare two assets,equities and bonds, and two shocks:a relativedemand shock for the goods producedby the home or foreign countries and a redistributiveshock that changes the accrualof income between laborand capitalwithin the country.The redistributiveshockexplainsthe home bias in equitiesbecausethe income risk to workersdue to this shock is hedged by holding claims on the return to domestic capital.In the state of the world where the redistributive shock transfersincome from labor to capital the workers are perfectly hedged as they still receive the same income, this time in the form of capitalincome due to theirdomestic equity holdings. The demand shock, on the other hand, explains the second stylized fact. The demand shock essentially redistributesincome across home and foreign countriesand having claims on the foreigncountryhedges this risk.An importantcontributionof the paper is showing thatin general a supply shock would not lead to the same predictiondue to terms of tradeeffects. Whilethe two shocksqualitativelyexplainthe two stylized factsin the completemarketscase, it is worthwhile to note thatthereis no interplay between these two in this case. The optimal portfolioholding response to the existence of one of the shocks (redistributive)explains the home bias in equities, while the response to the other (demand)explains the net long foreign currencyholdings. Thus, the complete markets case presents a model with two separatechannelsoperatingindependently, which is educative but is not completely satisfying. The model in this case also predictsextremeportfoliochoices such as completehome bias in equities. Tohave more interplaybetween the responses to the two shocks and to generate quantitativemodel predictions that are more in line with what is observedin the data,the authorsmove to an incompletemarkets settingby introducinga supply shock.Withtwo assets and threeshocks, marketsare incompleteand gross portfolioholdings are difficultto pin down. This difficulty is overcome with a very nice application of the Devereux-Sutherlandsolution method. The downside of the Devereux-Sutherlandsolution, which involves Comment 279 using a second-orderapproximation,is that it makes intuitionbuilding more difficult. While in the complete markets case it was clear which part of the model played which role in the results, this is no longer the case in the incompletemarketscase. In particular,what exactlyis the interplaybetween the threetypes of shocksthatlead to the particularportfolio holdings?Understandingthe natureof the interplaybetween these shocks and the portfolio holdings they give rise to will surely lead to more work in this field. In thatregard,this paperhas opened the door to a very interestingand potentially rewarding researchavenue in internationalfinance. Calculationof the LaborShare Only in the incomplete markets case can the model be sensibly calibrated as the complete marketsassumption leads to strong and counterfactualquantitativepredictions. The calibrationin the incomplete marketscase uses time variationin the labor share,calculatedfrom national statistics, to pin down the variance of the redistributiveshock. This may be problematicfor two reasons. Attributing the interpretationof a redistributive shock to annual changes in the labor share is similar to trying to measure fundamental totalfactorproductivity(TFP)changesfromannualSolow residuals.Although over long periods of time (e.g., five or ten years)this may be reasonable, there are too many measurement issues that complicate the analysis at higher frequencies. Changes in reporting practices, labor hoarding,and so forth,all affectthe measuredlaborshare,given thatthe variationin this is small to begin with, it is hard to be sure that not all of the annualvariationis due to measurementissues. A second, more fundamentalissue with the way this paper measures the laborshare is that it defines the laborincome as only compensation of employees. Thatis, LaborShare = Compensationof EmployeesI {GD?-IndirectTaxes). Buteven in industrializedcountriesnot all laboris employee labor.Part of the laborincome falls under the operatingsurplus of unincorporated enterprises(OSPUE)heading.As an example,table5C2.1shows the cost componentsof GDP in Italyin 1991. The operating surplus of unincorporatedenterprises in (4b) is the mom and pop stores' (small firms, as opposed to corporations)profits. As the owners'do not pay themselveswages, this proprietors'income is 280 Giirkaynak partly labor income. This is a nontrivialpart of the GDP- more than a quarterof the GDP is OSPUE- and not capturingthis will cause under measuringthe laborshare. This observation led Gollin (2002) and Bernanke and Giirkaynak (2001)to look for ways of dividing the OSPUEbetween laborand capital. One way of doing this is to assume the labor share in OSPUEis the same as in the rest of the economy.Anotherway is looking at the composition of the laborforce and inflatingthe employee compensationby the fractionof the laborforcethat are not employees. Table5C2.2shows the laborforce compositionin Italyin 1991. As canbe seen, employees makeup only about70 percentof the workforce,with the remaining30 percent'slaborincomebeing excludedfrom the labor share calculatedfrom employee compensationonly. The Italian labor share calculatedthis way is only 0.5, while it is a much more reasonable0.65 to 0.7 when correctedfor the labor income of the nonemployees. The currentpaper uses the standard deviation of the labor share to calibratethe variance of the redistributiveshock. Thus, mismeasuring its level may not be an issue by itself. However,both the shareof the opTable5C2.1 Cost componentsof GDP,Italy1991 1. Indirecttaxes,net 2. Consumptionof fixed capital 3. Compensationof employeesby residentproducers 4. Operatingsurplus 4a. Corporateand quasi-corporateenterprises 4b. Privateunincorporatedenterprises 4c. Generalgovernment 5. GrossDomesticProduct 133,361 168,539 647,792 477,879 71,312 403,714 2,853 1,427,570 Source:UN NationalAccountsStatistics Table5C2.2 Laborforcecomposition,Italy1991 Employersand own acct.workers Employees Unpaidfamilyworkers Not classifiableby status Total Source:ILOYearbookof LaborStatistics 5,228,000 15,478,000 886,000 2,653,000 24,245,000 International PortfolioswithSupply,Demand,andRedistributive Shocks 281 erating surplus in GDP and the compositionof the laborforceshow annual time variationof the same orderof magnitudeas the varianceof the laborsharecalculatedby the authors.Thatsuggests that calibratingthe variance of the redistributiveshock this way and from highly aggregated data, independently of whether such a shock is theoreticallyappealing or not, may not be very appropriate. Conclusion This paper identifies some of the most importantopen questions in the literatureand shows how new modeling devices can be used to provide answersto them.Whilethereareissues aboutthe modeling choices and calibrationpreferences, this way of thinking about the relevant questions will surely lead to more researchon these topics and a better understandingof internationalportfolio holdings. Researchthat especially asks how much of the observedportfoliochoices can be explained solely by hedging behavior, as in this paper (as opposed to informational and otherissues) would be most welcome. Such researchwill undoubtedlybenefitfromthe model and insights provided by this paper's authors. References Bernanke, B., and R. Gurkaynak. 2001. Is growth exogenous? Taking Mankiw, Romer, and Weil seriously. In NBERMacroeconomicsAnnual Vol. 16, 11-57. Cambridge, MA: MIT Press. Devereux, M., and A. Sutherland. 2006. Solving for country portfolios in open economy macro models. University of British Columbia. Working Paper. Gollin, D. 2002. Getting income shares right. Journalof Political Economy110 (2): 458-74.