Despite calls for its obsolescence, the lending channel of the monetary transmission mechanism continues to generate interest among monetary economists. Recently, Den Haan, Sumner, and Yamashiro (2007) find the loan-supply effect is evident for consumer lending and real estate loans in data up to 2004. The loan-supply effect, or lending channel, arises in the transmission mechanism if tight monetary policy forces banks to contract the supply of loans independent of a direct interest rate-effect (i.e., the "liquidity effect"), and such a contraction has real effects for bank-dependent borrowers. Although some, such as Perez (1998) and Ashcraft (2006), have emphasized the lending channel's lack of relevance for the monetary transmission mechanism, Den Haan, Sumner, and Yamashiro's (2007) results support a number of recent papers that offer detailed and disaggregated reasons why the channel is alive and well for certain lenders and borrowers (for recent examples, see Nilsen 2004 and Peek, Rosengren, and Tootell 2003). Den Haan, Sumner, and Yamashiro's (2007) findings suggest that the role of the consumer in the lending channel may be deserving of more attention. (1)
This paper asks whether consumer lending is, or has become, a significant component of the lending channel (even if it was not thought crucial in lending channel studies published in the 1990s). Specifically, I take a closer look at the statistical and possible economic significance of the consumer loan-supply effect with disaggregated monthly consumer credit data from both commercial bank and nonbank sources, and for the nonrevolving and revolving components of consumer credit from each source. With this array of data, I test the assumptions of the channel by considering whether households are forced to use more expensive nonbank credit as bank credit declines, or that a household might switch into more costly revolving credit as installment loans decline.
To examine the consumer lending channel, I estimate impulse response functions for the disaggregated consumer loan components mentioned above, with the monthly data spanning 1968-2006 (the sources of which are discussed later). This strategy follows the examples of numerous lending channel studies that analyze the responses of disaggregated data to a monetary policy shock to discern evidence of the loan-supply effect (Den Haan, Sumner, and Yamashiro 2007, e.g., compare the behavior of commercial and industrial loans with real estate and consumer loans). I follow the methods of previous studies in order to compare as closely as possible the results of this paper with the rich body of lending channel literature. For robustness, this paper estimates the impulse response functions from both a typical vector autoregression (VAR) model and using Jorda's (2005) linear projection technique. The latter method provides impulse response functions less-prone to misspecification (than in a VAR) along with conditional standard errors to aid in the statistical inference of the responses (see also Jorda 2009 and further discussion in Section III). Finally, for additional robustness, I employ Den Haan, Sumner, and Yamashiro's (2007) identification strategy for isolating the loan-supply effect.
In preview of the results, ultimately the consumer credit data analyzed in this paper suggest that both the statistical (and by extension, the economic) significance of the consumer loan-supply effect is weak; in the least, the effect has weakened over time. Although monthly consumer credit data from commercial banks from 1968 through 2006 substantiate previous research on consumer lending--matching the decline documented by Gertler and Gilchrist (1993) and Den Haan, Sumner, and Yamashiro (2007)--after 1984, the responses of both nonrevolving and revolving consumer bank credit are not consistent with the lending channel. Instead, both series increase, similar to the result found by Den Haan, Sumner, and Yamashiro (2007) for commercial lending (in fact, total nonrevolving and revolving consumer credit increase for 2 yr after the monetary shock). …