چرا دسترسی شرکتها برای بازار اوراق قرضه تفاوت زیادی با دسترسی به چرخه کسب و کار دارد؟ یک نظریه و شواهد

This paper presents a theory of firm access to the bond market in which information gathering agencies are valuable but alter the relative cost of bond financing across firms and over the business cycle. The theory builds on the assumption that information frictions prevent these agencies from rating firms correctly all of the time. As a result, the cost of bond financing becomes dependent on the state of the economy and the “quality” of the signal provided by these agencies’ ratings. In addition, when the mix of bond issuers becomes riskier, as happens in recessions, bond financing becomes more expensive for mid-quality firms. Bond financing may even become more expensive to all firms, in which case mid-quality firms will be affected the most. The analysis of the bonds issued in the last two decades by American firms shows that split ratings, our proxy for the “quality” of the rating agencies’ signal, do not affect the relative cost of bond financing across firms in expansions, but they do increase the relative cost of this funding source for mid-credit quality issuers in recessions.

Evidence from the US bond market shows that recessions increase bond credit spreads in the primary market, affecting the relative cost of bond financing across issuers of different creditworthiness. In this paper, we argue, and present supporting evidence, that the effect of recessions on the relative cost of bond financing is partly due to the information agencies that firms use to raise bond financing. Our theory builds on the assumption that information frictions prevent these agencies from rating firms correctly all of the time. As a result, the cost of bond financing becomes dependent on the state of the economy and the “quality” of the signal provided by these agencies’ ratings. There are many theories that explain firms’ use of bond financing, but surprisingly none of these theories take into account the role of rating agencies.1 Yet, virtually all firms use these agencies to access the bond market. The widespread use of credit ratings, of course, could be due to institutional factors like regulations, but it could also be attributable to a valuable role performed by rating agencies.2Ramakrishnan and Thakor, 1984, Millon and Thakor, 1984 and Boot et al., forthcoming, for example, provide theories that explain intermediaries whose main function is to produce information to be used by investors. Liu and Thakor, 1984, Ederington et al., 1987 and Hand et al., 1992, in turn, provide evidence that rating agencies produce valuable information. This paper adds to the literature on bond financing by presenting a theory of firms’ access to the bond market in which rating agencies provide a valuable service. We create a role for these agencies by assuming that there is adverse selection. This makes it worthwhile for firms to contract with an information agency in exchange for a signal on their creditworthiness. Importantly, the same asymmetry of information that makes these agencies valuable also leads them to produce occasionally incorrect assessments about the firm’s creditworthiness. Following the idea that it is relatively easier for information agencies to identify the worst and best firms in the economy, we assume that the “quality” of the signal produced by these agencies is lower for mid-credit quality firms. Under these conditions, we show that the cost of accessing the bond market depends on the “quality” of the signal of information agencies’ ratings. We further show that if distribution of firms is riskier in recessions (in a first order stochastic dominance sense), then recessions will increase the cost of information agencies’ incorrect assessments. As a result, and for this reason alone, recessions increase the cost of bond financing to mid-quality firms. Recessions may also increase the cost of bond financing to all firms, in which case mid-quality firms are affected the most. To test our theory we first try to find supporting evidence for the model’s assumption on the “quality” of the signal produced by information agencies. We proxy the “quality” of this signal by the frequency of bond rating splits at issue date between Moody’s and S&P under the assumption that there is an inverse relationship between the “quality” of the signal of these agencies’ ratings and the frequency with which they disagree on the ratings they assign bonds. Our finding that rating agencies are more likely to announce split ratings on bonds issued by mid-credit quality firms is consistent with our assumption that the “quality” of the ratings’ signal is lower for these firms. We then investigate bond credit spreads at issue date in an attempt to find supporting-evidence for the model’s implications on the role of rating agencies on the cost of bond financing. Consistent with our theory, the results of this analysis show that, ceteris paribus, rating splits do not alter the relative cost of bond financing across firms in expansions but they do so in recessions. On these periods, they increase the cost of bond financing, and their effect is largest for mid-credit quality issuers. Our empirical analysis is related to Fama and French, 1989 and Bernanke, 1993, who also show that recessions increase the cost of bond funding. However, in contrast to Bernanke (1993), who argues that the widening of spreads in downturns is driven by investors demand, our paper provides evidence that some of that increase in the spreads is due to the role that rating agencies play when firms raise bond financing. Our empirical analysis also adds to the literature on bond pricing in the primary market by investigating the impact of the “quality” of the signal produced by rating agencies on the cost of this funding source. Cantor et al., 1997 and Jewell and Livingston, 1998 investigate the impact of split ratings on bond spreads, but they do not control for the state of the economy at the time of the issue. The remainder of the paper is organized as follows. The next section presents our theory and derives some empirical implications. Section 3 introduces the data we use to test these implications. Section 4 tests the model’s key assumption on the “quality” of the signal produced by information agencies, and Section 5 tests the model’s implications on the impact of the “quality” of this signal on the cost of bond financing. Section 6 concludes the paper.

Our paper presents a theory of firm access to the bond market in which information agencies provide a valuable service, but alter the relative cost of accessing this market across firms of different creditworthiness and over the business cycle. Our theory explains how the state of the economy can influence the cost of bond financing under the assumption that firms fund their investments solely with bond financing. Williamson, 1987, Bernanke and Gertler, 1989 and Rajan, 1994 present theories linking bank credit policies to the state of the economy in settings where firms fund themselves exclusively with bank loans. Given that firms with access to the bond market often continue to use bank funding, merging these two strands of the literature to investigate how firms alter their relative use of both funding sources over the business cycle seems a fruitful area for future research.