Empirical analysis of corn and soybean basis in Canada

Abstract

The article examines the factors affecting the basis for corn and soybeans using several time-series techniques to account for potential structural breaks, seasonality, residual serial correlation and structural breaks, as well as potential endogeneity and nonstationarity. The spatio-temporal empirical framework is based on storage and trade theories which assume the relationship between nondelivery location’s spot price and futures price of a storable commodity depends on opportunity cost of capital, warehousing costs, a convenience yield and shipping costs. The interest rate effect is strong for both crops with shipping costs also affecting soybean basis and own inventory levels positively correlated with corn basis. The effect of the wedge between the price of carrying physical grain and the maximum storage rate on basis is positive for both crops. The empirical results, which are robust to multiple estimators, provide stronger evidence of a structural break for the soybean basis than for the corn basis.

Keywords:

• nondelivery point basis
• structural breaks
• wedge
• nonconvergence
• time series

JEL Classification:

• Q13
• G13
• C22
• C52

Acknowledgements

We thank an anonymous referee for insightful comments.

Notes

¹ See Inoue and Hamori (2014) for recent application of FMOLS and DOLS.

² We set the trimming parameter to 0.10 T, where T is the sample size.

³ The Schwartz modified criterion of Liu et al. (1997) and the BIC select zero and one break, respectively, for soybean (and no breaks for corn). The results for model selected for BIC with one break are available upon request. The sign and significance of the primary explanatory variables remain unchanged relative to those shown Table 9 with two breaks, except for the coefficient on transportation, which becomes negative and insignificant. However, Bai and Perron argue that their sequential procedure has better properties for selecting the number of breaks.

⁴ We use the maximum lag rule L = integer(12(T/100)¹/⁴) (Hayashi, 2000, p. 594). The rule indicates a maximum lag of 14.

⁵ The wedge contains a future difference in the discounted basis. Although the relationship between a difference and a level is often weak, we cannot completely rule out the possible endogeneity of the wedge in the equation for the basis. In additional results available upon request, we considered (twice) lagged wedge as a possible instrument for wedge in both the corn and soybean specifications. We also considered stock as an instrument for the soybean specification since the results of Table 13 suggest that stock can be excluded from the soybean regression given the inclusion of wedge. However, we could not reject the null hypothesis that these instruments were weak employing either the Stock and Staiger (1997) rule of thumb or the Stock and Yogo (2005) weak instrument test. The construction of a stronger instrument for wedge and a fuller examination of the relationship between the wedge and basis could be an interesting topic for future research.

Additional information

Funding

Maynard thanks the SSHRC for research funding. This work was supported by the Ontario Ministry of Agriculture, Food and Rural Affairs [grant number 200289].