Practitioners' Corner: Introduction to the Special Issue

The first 150 words of the full text of this article appear below. The material in this volume is the result of a call for papersto the participants of the "Conference on Analysis of High-FrequencyFinancial Data and Market Microstructure" held in December 2003in Taipei, Taiwan. Jeffrey Russell and Ruey Tsay have actedas guest editors for this special issue, together with the editorsRené Garcia and Eric Renault. The availability of high-frequency data has spawned considerableliterature on volatility measurement and forecasting. The materialis mathematically delicate and perhaps "Practitioners’Corner" would be well advised to let the dust settle a bit tosee what emerges at the end of the day. On the other hand, thepractically minded may well be served by a good road map ofthe issues. So with only mild apology do we take up the cartographyof some difficult terrain. To fix ideas, let S(t) denote the price process of a . . . [Full Text of this Article]     

Practitioners' Corner

The first 150 words of the full text of this article appear below. Spring at last. Here in Montreal, the transition from winteris particularly delicious. There is more light, the world looksless grey, and the trepidation that another storm may be aroundthe corner is put away with the winter tires. In this issueof JFEC, the editors appear to be underscoring seasonal changein an issue whose unifying theme involves parameter and modelstability. Parameter stability is an important issue in econometrics. Achanging economic environment may be captured by allowing theparameters of a reduced form model to vary to reflect changingconditions. The challenge for the econometrician is to constructmodels that do more than reflect changes in an ad hoc manner.For want of more insightful approaches, the challenge is oftensidestepped in practice by focusing on shorter samples whereno structural change can be assumed to have occurred. However,in so constricting the sample size, . . . [Full Text of this Article]     

Practitioners' Corner

The first 150 words of the full text of this article appear below. The analysis of volatility remains a preoccupation. In our veryfirst issue, "Practitioners' Corner" offered a brief retrospectiveon volatility modeling, surveying several strategies in thehistory of volatility modeling and locating the contributionsof the first issue within these broad themes. Of course, notall the highlights of this voluminous literature could be visitedor all noteworthy references cited. Nonetheless, we should havereferred to the venerable literature on mixture models introducedby the polymath Simon Newcomb in the late 19th century and subsequentlystudied by Karl Pearson. A neglected reminder was certainlysupplied by Lanne and Saikkonen (2003), who in this same firstissue of JFEC offered the wry understatement that the conditionalheteroskedasticity inherent in mixture autoregressive modelsmay not adequately capture the time-series properties of financialdata. The point is made again in the contribution to this issueby Markus Haas, Stefan Mittnik, and Marc . . . [Full Text of this Article]     

Practitioners' Corner

The first 150 words of the full text of this article appear below. This issue of the Journal of Financial Econometrics containspapers that were presented at the (EC)² conference Econometricsof Financial and Insurance Risks held in Istanbul on December15–17, 2005. Launched in 1990, (EC)² is an annual seriesof international conferences on research in quantitative economicsand econometrics. The acronym stands for European Conferencesof the Econom[etr]ics Community. Its primary aim is to providea vibrant forum where both senior and junior European researchersin quantitative economics and econometrics can discuss the resultsand progress of their research. (EC)² conferences are of relativelysmall scale (less than 200 participants) and quite intensive.Each year a different topic is selected as the major theme ofthe conference. A few leading quantitative economists or econometriciansare invited as keynote speakers, such as Eric Ghysels, who alsoacts as co-Guest editor of this issue; the other speakers areselected . . . [Full Text of this Article]     

Practitioners' Corner

The first 150 words of the full text of this article appear below. The transition from some preliminary qualitative assessmentsto quantitative assertions is a hallmark of scientific progress.One dollar today is more valuable than one dollar availabletomorrow, but the scientific issue is by how much and why thisamount? Similarly a random gain with unit expectation and significantrisk is less valuable than one dollar available with certainty.Financial theories of the discount rate and the mean-variancetrade-off have afforded helpful quantitative answers to thesecrucial issues in asset pricing, investment, and risk management.Meanwhile, modern financial econometrics has characterized thedynamic features of interest rates as well as of risk and returnthrough state variables models with volatility clustering, mean-volatilityfeedback, and dynamic correlations. From this perspective, a striking common feature of all thearticles in the current issue of the Journal of Financial Econometricsis the reintroduction of qualitative variables previously treatedusing purely quantitative approaches. The first . . . [Full Text of this Article]     

Practitioners' Corner

The first 150 words of the full text of this article appear below. In our discussion in the last issue of Journal of FinancialEconometrics (JFEC) of the nonparametric methods developed byBarndorff-Nielsen and Shephard (2006) to detect jumps in thelocal behavior of the continuous time path of a price process,we observed these tests were not designed to detect major pricediscontinuity events such as the 1987 crash, since the testingmethodology precludes jumps in adjacent time intervals. Indeed,a major event such as Black Monday is characterized by a sequenceof jumps in consecutive time intervals throughout the day. Inthe interest of thematic continuity, let’s pursue thematter of jumps further. The first article in the current issue by Hossein Asgharianand Chistoffer Bengtsson addresses directly the detection ofbig events in stock prices. More particularly, the authors analyzethe spillover of jumps across international stock markets. Tomeasure jumps, the authors formulate a parametric model in . . . [Full Text of this Article]     

Practitioners' Corner

The first 150 words of the full text of this article appear below. During the 1980s, the early stages of modeling financial timeseries focused on the striking stylized fact that while returnswere themselves not serially correlated, squared returns were.This history has been nicely documented in the influential bookby Taylor (1986) and, indeed, the opening chapters of contemporaryfinancial econometrics open with Engle (1982) and Bollerslev(1986) who provided a specific ARMA structure of squared returnsvia the celebrated [G]ARCH models. This general orientationin effect acknowledged that there was some room for predictingrisk, as measured by squared values or absolute values of returns,while at the same time maintaining the hypothesis that returnsthemselves were hardly predictable in keeping with some versionof market efficiency. However, this paradigmatic view has beenchallenged over the subsequent 20 years in at least three regards. First, with Nelson (1991), it has been widely acknowledged thatalthough GARCH modeling is about . . . [Full Text of this Article]     

Practitioners' Corner

The first 150 words of the full text of this article appear below. The article by Christian Gourieroux and Razvan Sufana providesa complete characterization of two-factor affine diffusion termstructure models. The presentation of the article may seem forbiddingto the practitioner audience, so it is perhaps useful that thisinstallment of the column provide some context for the resultsof this article. Prominent among affine diffusion models in the term structureliterature are the Gaussian and square-root diffusion modelsof Vacisek (1977) and Cox, Ingersoll, and Ross (1985). The seminalcontribution of Duffie and Kan (1996) had been to provide anecessary and sufficient condition on the stochastic model toobtain the desirable property of "affine yields," whereby theyield of any zero-coupon bond is seen as a maturity-dependentaffine combination of a selected "basis" set of yields. SubsequentlyDuffie, Pan, and Singleton (2000) proved that the aforementionedcharacterization is even more general, both for the stochasticmodel that may . . . [Full Text of this Article]     

Inequality Constraints in the Fractionally Integrated GARCH Model

In this article we derive necessary and sufficient conditionsfor the nonnegativity of the conditional variance in the fractionallyintegrated generalized autoregressive conditional heteroskedastic(p, d, q) (FIGARCH) model of the order p 2 and sufficient conditionsfor the general model. These conditions can be seen as beinganalogous to those derived by Nelson and Cao (1992, Journalof Business & Economic Statistics 10, 229–235) forthe GARCH(p, q) model. However, the inequality constraints whichwe derive for the FIGARCH model illustrate two remarkable propertiesof the FIGARCH model which are in contrast to the GARCH model:(i) even if all parameters are nonnegative, the conditionalvariance can become negative and (ii) even if all parametersare negative (apart from d), the conditional variance can benonnegative almost surely. In particular, the conditions forthe (1, d, 1) model substantially enlarge the sufficient parameterset provided by Bollerslev and Mikkelsen (1996, Journal of Econometrics73, 151–184). The importance of the result is illustratedin an empirical application of the FIGARCH(1, d, 1) model toJapanese yen versus U.S. dollar exchange rate data.     

The BDS Test as a Test for the Adequacy of a GARCH(1,1) Specification: A Monte Carlo Study

In this study we examine the widely used Brock, Dechert, andScheinkman (BDS) test when applied to the logarithm of the squaredstandardized residuals of an estimated GARCH(1,1) model as atest for the adequacy of this specification. We review the conditionsderived by De Lima (1996; Econometric Reviews 15, 237–259)for the nuisance-parameter-free property to hold and addressthe issue of their necessity, using the flexible framework offeredby the GARCH(1,1) model in terms of moment, memory, and timeheterogeneity properties. By means of Monte Carlo simulations,we show that the BDS test statistic still approximates the standardnull distribution even for mildly explosive processes that violatethe majority of the conditions. Thus the test performs reasonablywell, its empirical size being rather close to the nominal one.As a by-product of this study, we also shed light on the relatedissue of the consistency of the QML estimators of the conditionalvariance parameters under various parameter configurations andalternative distributional assumptions on the innovation process.     

Stochastic Volatility With an Ornstein-Uhlenbeck Process: An Extension

In this paper, we reexamine and extend the stochastic volatilitymodel of Stein and Stein (S&S) (1991) where volatility followsa mean–reverting Ornstein–Uhlenbeck process. UsingFourier inversion techniques we are able to allow for correlationbetween instantaneous volatilities and the underlying stockreturns. A closed-form pricing solution for European optionsis derived and some numerical examples are given. In addition,we discuss the boundary behaviour of the instantaneous volatilityat v(t) = 0 and show that S&S do not work with an absolutevalue process of volatility. JEL Classification: G13     

LARCH, Leverage, and Long Memory

We consider the long-memory and leverage properties of a modelfor the conditional variance of an observable stationary sequence X_t, where is the square of an inhomogeneous linear combination of X_s, s < t, withsquare summable weights b_j. This model, which we call linearautoregressive conditionally heteroskedastic (LARCH), specializes,when depends only on X_t–1, to theasymmetric ARCH model of Engle (1990, Review of Financial Studies3, 103–106), and, when depends only on finitely many X_s, to a version of the quadratic ARCH modelof Sentana (1995, Review of Economic Studies 62, 639–661),these authors having discussed leverage potential in such models.The model that we consider was suggested by Robinson (1991,Journal of Econometrics 47, 67–84), for use as a possiblylong-memory conditionally heteroskedastic alternative to i.i.d.behavior, and further studied by Giraitis, Robinson and Surgailis(2000, Annals of Applied Probability 10, 1002–1004), whoshowed that integer powers , =" BORDER="0">2 can have long-memory autocorrelations. We establish conditionsunder which the cross-autocovariance function between volatilityand levels, , decays in the manner of moving average weights of long-memory processes on suitable choiceof the b_j. We also establish the leverage property that h_t <0 for 0 < t k, where the value of k (which may be infinite)again depends on the b_j. Conditions for finiteness of thirdand higher moments of X_t are also established.     

Nondisclosure and adverse disclosure as signals of firm value

We present a model in which some of the firm's information ('news')can be disclosed verifiably and some information ('type') cannot,to show that some firms may voluntarily withhold good news anddisclose bad news. We describe an equilibrium in which high-typefirms withhold good news and disclose bad news, whereas low-typefirms disclose good news and withhold bad news. Under some parametervalues, this equilibrium exists when other more traditionalequilibria are ruled out by standard equilibrium refinements.The model explains some otherwise anomalous empirical evidenceconcerning stock price reactions to disclosure, provides somenew empirical predictions, and suggests that mandatory disclosurerequirements may have the undesirable consequence of makingit more difficult for firms to reveal information that cannotbe disclosed credibly.     

Household Portfolio Diversification: A Case for Rank-Dependent Preferences

The proliferation of novel preference theories in financialeconomics is hampered by a lack of non-experimental evidenceand by the theories’ additional complexity which has notbeen shown to be critical in applications. In this article Ipresent arguments in support of preferences with rank dependency.Using the Survey of Consumer Finances data, I document two widespreadpatterns inconsistent with expected utility: (i) many householdssimultaneously invest in well-deversified funds and in poorly-diversifiedportfolios of stocks; and (ii) some households with substantialsavings do not invest anything in equities. I show that portfoliochoice models with rank-dependent preferences, plausibly parameterizedand under fully rational assumptions, are quantitatively consistentwith the observed diversification. These results call for furtherefforts to integrate the models of rank-dependent preferencesin portfolio theory and asset pricing.     

Practitioners' Corner: Introduction to the Special Issue

This issue of the Journal of Financial Econometrics comprisespapers that were presented at the workshop "New Directions inFinancial Risk Management," held under the aegis of the Centerfor Financial Studies in Frankfurt on November 3–4, 2003.The scientific organizers of the conference were Frank     

Theory and Evidence on the Resolution of Financial Distress

We analyze a financially distressed owner-managed project. Themain results of the model are: (1) borrower default is an endogenousresponse to the anticipated restructuring–foreclosureoutcome; (2) the lender’s restructuring–foreclosuredecision depends critically on the interaction between projectvalue and industry liquidity; and (3) the lender waits for theindustry to recapitalize before selling assets obtained throughforeclosure. Empirical analysis of a large sample of defaultedcommercial real estate loans supports many of the model predictions,including restructuring–foreclosure outcomes that areconsistent with endogenous borrower default and firesale discountsthat vary depending on industry market conditions at the timeof foreclosure. (JEL G33)     

Auctions vs. Bookbuilding and the Control of Underpricing in Hot IPO Markets

Market returns before the offer price is set affect the amountand variability of initial public offering (IPO) underpricing.Thus an important question is "What IPO procedure is best adaptedfor controlling underpricing in "hot" versus "cold" market conditions?"The French stock market offers a unique arena for empiricalresearch on this topic, since three substantially differentissuing mechanisms (auctions, bookbuilding, and fixed price)are used there. Using 1992–1998 data, we find that theauction mechanism is associated with less underpricing and lowervariance of underpricing. We show that the auction procedure'sability to incorporate more information from recent market conditionsinto the IPO price is an important reason.     

A Mixture Multiplicative Error Model for Realized Volatility

A multiplicative error model with time-varying parameters andan error term following a mixture of gamma distributions isintroduced. The model is fitted to the daily realized volatilityseries of deutschemark/dollar and yen/dollar returns and isshown to capture the conditional distribution of these variablesbetter than the commonly used autoregressive fractionally integratedmoving average model. The forecasting performance of the newmodel is found to be, in general, superior to that of the setof volatility models recently considered by Andersen et al.(2003, Econometrica 71, 579–625) for the same data.     

Walrasian Tatonnement Auctions on the Tokyo Grain Exchange

The Tokyo Grain Exchange (TGE)’s itayose mechanism providesthe opportunity to analyze functioning Walrasian tâtonnementauctions (WTA). In 15,677 auctions conducted over 1997–1998for corn and redbean futures contracts, price formation is unexpectedlysimilar to that observed in continuous double auctions. Provisionalprices and pledges are informative. In contrast to behaviorobserved in experiments, few pledges are deceptive, becausethe traders participate repeatedly and because the auctioneerhas flexibility when changing the provisional price and endingthe auction. Both the risk of the auction ending and the moreequitable dispersion of information increase depth and the speedat which information is embodied in price.     

Model-free versus Model-based Volatility Prediction

The well-known ARCH/GARCH models for financial time series havebeen criticized of late for their poor performance in volatilityprediction, that is, prediction of squared returns.¹ Focusingon three representative data series, namely a foreign exchangeseries (Yen vs. Dollar), a stock index series (the S&P500index), and a stock price series (IBM), the case is made thatfinancial returns may not possess a finite fourth moment. Takingthis into account, we show how and why ARCH/GARCH models—whenproperly applied and evaluated—actually do have nontrivialpredictive validity for volatility. Furthermore, we show howa simple model-free variation on the ARCH theme can performeven better in that respect. The model-free approach is basedon a novel normalizing and variance–stabilizing transformation(NoVaS, for short) that can be seen as an alternative to parametricmodeling. Properties of this transformation are discussed, andpractical algorithms for optimizing it are given.