commodity market capital flow and asset return yogo.pdf · commodity market capital flow and asset...

Download Commodity Market Capital Flow and Asset Return Yogo.pdf · Commodity Market Capital Flow and Asset Return ... To formalize our observation, we regress the monthly excess returns on

Post on 30-Jan-2018

212 views

Category:

Documents

0 download

Embed Size (px)

TRANSCRIPT

  • Commodity Market Capital Flow and Asset Return

    Predictability

    Harrison Hong Motohiro Yogo

    February 28, 2010

    Abstract

    We establish several new findings on the relation between capital flow in commod-ity markets and asset returns. Capital flowing into commodity markets, as measuredby high open-interest growth, predicts high commodity returns and low bond returns.Open-interest growth is a more powerful and robust predictor of commodity returnsthan other known predictors such as the short rate, the yield spread, the basis, andhedging pressure. It is positively correlated with commodity returns but has informa-tion for future returns beyond that contained in past commodity prices. Open-interestgrowth also predicts changes in inflation and inflation expectations. These findingssuggest that open-interest growth contains information about future inflation that getspriced into commodity and bond markets with delay. Our findings are consistent withrecent theories of gradual information diffusion and have implications for macroeco-nomic forecasting models.

    This paper subsumes our earlier work titled Digging into Commodities. For comments and discussions,we thank Erkko Etula, Hong Liu, David Robinson, Nikolai Roussanov, Allan Timmermann, and seminarparticipants at Boston College, Centre de Recherche en Economie et Statistique, Dartmouth College, Ford-ham University, PanAgora Asset Management, Stockholm School of Economics, University of California SanDiego, University of Pennsylvania, University of Southern California, Washington University in St. Louis,the 2008 Economic Research Initiatives at Duke Conference on Identification Issues in Economics, and the2010 Annual Meeting of the American Finance Association. We thank Jennifer Kwok, Hui Fang, YupengLiu, James Luo, Thien Nguyen, and Elizabeth So for research assistance. Hong acknowledges a grant fromthe National Science Foundation. Yogo acknowledges a grant from the Rodney L. White Center for FinancialResearch at the University of Pennsylvania.

    Princeton University and NBER (e-mail: hhong@princeton.edu)University of Pennsylvania and NBER (e-mail: yogo@wharton.upenn.edu)

  • 1. Introduction

    We analyze how capital flow in commodity markets is related to commodity and bond re-

    turns. Our analysis is motivated by the recent volatility in commodity prices and the renewed

    interest in the behavior of these markets, which have not been seen since the energy crisis of

    the 1970s. Once largely ignored by the investment community, commodities have emerged as

    an important asset class. By some estimates, index investment in this asset class increased

    from $13 billion at the end of 2003 to $317 billion in July 2008, just prior to the financial

    crisis (Masters and White, 2008). During the same period, the influx of new investors led to

    elevated levels of capital flow as measured by open interest in commodity futures, which grew

    from $103 billion to $509 billion. This capital flow has led to inquiries about how trading

    affects asset price formation in these markets, underscored by the recent Congressional hear-

    ings on the impact of excessive speculation. Hence, understanding the link between capital

    flow and commodity price fluctuations is important not only for investors but also for public

    policy.

    Our analysis covers 30 commodities across four sectors (agriculture, energy, livestock,

    and metals) over the period of 1965 through 2008. Using hand-collected data on open inter-

    est from the Commitments of Traders since 1965, we establish several new findings on the

    relation between capital flow and returns in commodity markets. Figure 1 summarizes our

    main finding. The first series is the percentage change in open interest over the previous

    12 months, averaged across all commodities. The second series is the return on fully collat-

    eralized commodity futures over the previous 12 months, averaged across all commodities.

    During the recent commodity boom from 2003 to 2008, capital flowed into the sector at a

    persistently high rate, more so than in any other period over the previous thirty years. Only

    the energy crisis of the 1970s witnessed higher activity. During these two historic periods

    and also more generally, open-interest growth and commodity returns are highly correlated.

    But the most interesting finding in this plot is that open-interest growth seems to lead com-

    modity returns. In other words, capital flowing into commodity markets appears to predict

    2

  • subsequent appreciation of commodity prices.

    To formalize our observation, we regress the monthly excess returns on a portfolio of

    commodity futures onto lagged 12-month open-interest growth. We find that a standard

    deviation increase in open-interest growth increases expected commodity returns by 0.64%

    per month. Similarly, we find that a standard deviation increase in open-interest growth

    increases expected spot-price growth by 0.41% per month. Both of these estimates are

    economically large and statistically significant. Open-interest growth is a more powerful and

    robust predictor than a number of other variables that are known to predict commodity

    returns. These include common predictors such as the short rate and the yield spread and

    commodity-specific predictors such as aggregate basis (i.e., the ratio of futures to spot price

    averaged across commodities) and aggregate hedging pressure (i.e., the net short position

    of hedgers averaged across commodities).1 Open-interest growth is a more robust predictor

    than these other variables in two important ways. First, aggregate open-interest growth

    predicts returns on sector portfolios, in contrast to other variables that predict returns for

    only particular sectors. Second, open-interest growth is the only variable that continues

    to demonstrate forecasting power in the most recent period since 1987, when there are the

    greatest number of commodities in the database.

    Open-interest growth is most closely related to 12-month commodity returns. We find

    that past aggregate commodity returns forecast the subsequent months return. In other

    words, there is momentum in the time series of aggregate commodity returns. However, in a

    horse race between these variables, open-interest growth entirely drives out the forecasting

    power of past commodity returns. This means that open-interest growth contains informa-

    tion about future returns that is not fully captured by past commodity prices. A potential

    1Bessembinder and Chan (1992) are the first to establish that the same variables that predict bond andstock returns (such as the short rate, the default spread, and the dividend yield) also predict commodityreturns. There is mixed evidence that basis predicts returns on commodity futures. Fama and French (1987)are the first to establish that basis predicts returns for some commodities. They emphasize that there ismore consistent evidence for the theory of storage. A number of other studies have documented mixedevidence for the theory of backwardation, controlling for systematic risk and using an empirical proxy forhedging pressure (Carter, Rausser, and Schmitz, 1983; Chang, 1985; Bessembinder, 1992; de Roon, Nijman,and Veld, 2000).

    3

  • interpretation of these findings is that capital flows into commodity markets in response to

    news, which get impounded into commodity prices with delay. In the 1970s, for example,

    there were news about supply shocks to oil. In the most recent period, there were news

    about strong demand for commodities from the emerging economies.

    To test this hypothesis, we examine whether open-interest growth predicts inflation and

    excess bond returns. Consistent with the hypothesis, we find that high open-interest growth

    predicts rising inflation and also a rising nominal short rate. In addition, high open-interest

    growth predicts low bond returns with a t-statistic over 3. A standard deviation increase in

    open-interest growth decreases expected bond returns by 0.32% per month. Open-interest

    growth is the only predictor that survives in the most recent period since 1987, when the

    short rate and the yield spread fail to predict bond returns. Hence, open-interest growth

    not only contains powerful information about future commodity returns, but also important

    information about future bond returns.

    To summarize, our novel finding is that capital flow into the commodity sector contains

    information about inflation news and bond returns that is not fully captured by commodity

    prices. As we discuss in the body, our findings are most consistent with recent theories

    of gradual information diffusion in asset markets (see Hong and Stein, 2007, for a review).

    These theories suggest that when market prices under-react to news, trading activity emerges

    as a useful additional predictor of future returns. Moreover, the fact that capital flow can be

    useful for predicting economic activity like inflation expectations has important implications

    for macroeconomic forecasting models.

    Our work is part of a second generation of commodity papers that have recently emerged

    in response to renewed interest in commodity markets. In a pioneering study that lays

    out the agenda, Gorton and Rouwenhorst (2006) emphasize that commodities have a high

    Sharpe ratio and a low correlation with other asset classes. They argue that this evidence

    is consistent with the theory of backwardation in particular and market segmentation more

    generally. Acharya, Lochstoer, and Ramadorai (2009) f

Recommended

View more >