The Effect of firms’ MBE recurrence on loan spread and maturity
Another stream of empirical research finds that firms with patterns of increasing earnings have higher price-earnings multipliers than others (Barth et al. 1999). Others further demonstrate that the market premium is more pronounced for firms that consistently beat earnings expectations (“habitual beaters”) than for only occasionally
22
do so (“sporadic beaters”) (Bartov et al. 2002; Kasznik and McNichols 2002). These findings are in line with the “momentum story”.
In this section, we further shed light on the effect of MBE recurrence on loan terms in the case of private debt market. In doing so, we construct one proxy for MBE recurrence (FMBE4), measured as the frequency of achieving MBE during the consecutive four quarters immediately before the matching fiscal quarter for a loan initiation on date t. To examine this prediction, we restrict our sample to firms meeting earnings expectations.14 The results are reported in Table 6.
As shown in column (1) of Table 6, the coefficient of FMBE4 is significantly negative, suggesting that loan spread decreases with frequency of MBE. Similarly, as shown in column (3) of Table 6, the coefficient of FMBE4 is significantly positive in the sense that loan maturity increases with frequency of MBE. Overall, the benefits of MBE in terms of loan spread and maturity increase with MBE frequency.
Next, we examine whether the benefits of habitual MBE are relatively greater for firms in financial distress. As evidenced by column (2) of Table 6, the coefficient of FMBE4*DISTRESS is significantly negative, as predicted. In addition, column (4) of Table 6 shows that the coefficient of FMBE4*DISTRESS is positive but insignificant. Alternatively, we follow Bartov et al. (2002) and construct a dichotomous indicator variable HMB4 to identify firms as being ‘‘sporadic beaters’’, based on the criterion that firms meeting or beating earnings expectations in at least 75% of the previous four quarters. Untabulated results are the same as those in Table 6.
Overall, these results lend support to the argument that the benefits of habitual
14 Our results are robust when we use the full sample to test our predictions.
23
or sporadic MBE in terms of loan spread are relatively greater for financially distress firms than for financially sound firms. However, this inference is slightly weaker in the case of loan maturity.
[INSERT TABLE 6 HERE]
Can Prospect Theory account for our findings?
As mentioned earlier, two theories are proposed to account for the MBE:
transaction cost theory and prospect theory. Due to their superior access to the information from the borrowers, private lenders such as banks are unlikely to rely on earnings information heuristics or reference points to evaluate borrowers’ credit worthiness’ credit worthiness when designing loan contract terms. Consequently, we conjecture that the results in this paper should be primarily accounted for by the prospect theory, rather than transaction cost theory. In this section, we will take direct examination of the association between loan contract terms and MBE from the perspective of prospect theory.
Prospect theory postulates that decision-maker derives value from gain and loss with respect to some reference points, and also suggests that value function is steepest around the wealth reference points. Thus, the value function is S-shaped with the curve concave for gains (implying risk-averse) and convex for losses (implying risk-seeking), and the corresponding losses is steeper than that for gains. As indicated by Degeorge et al. (1999), meeting analysts’ earnings forecasts is an important reference point. We might be able to anticipate that lenders’ risk premium depends on the loan spread they are willing to pay.
In this setting, the lender derives value from gain and loss with respect to analysts’ earnings forecast. Specifically, unexpected earnings (e.g., the difference
24
between actual earnings and analysts’ forecast earnings) is a natural reference point for lenders who estimate wealth as a multiple of loan spread which is negatively related to earnings. Assuming that firms manipulate wealth measures (changes in earnings and in turn changes in spreads) to affect the value perceived by lenders, we expect to observe earnings-increasing management around wealth reference points－
in our setting, in the vicinity of zero changes of unexpected earnings. The above arguments can be summarized in Panel (a) of Figure 2.
Due to the negative relation between loan spread and actual earnings, we thus multiply Loan Spread by negative one to convert it into S-shaped value function as postulated in prospect theory (See Panel (b) of Figure 2).
[INSERT FIGURE 2 HERE]
On the other hand, as noted in Gottesman and Roberts (2004), lenders may be willing to offer long-term or short-term loans to riskier borrowers at higher spreads at the same time. Therefore, the association between earnings and the length of loan maturity is not clear. In our context, we examine the benefits of MBE (i.e., lower loan spread) and spread-earnings association under the prospect theory framework;
specifically, we examine whether the MBE benefits are greater for losses than for gains when earnings move from losses to gains relative to analysts’ earnings forecasts.
To test our conjecture, we use the negative of loan spread, denoted Loan
Spread*(-1), as dependent variable and run the following proposed quadratic
regressions for firms meeting or beating earnings expectations and those missing earnings expectations, respectively:25 where Earnings is defined as the Compustat actual earnings; all other variables are as previously defined. We use
1MBE 2
2MBEEarnings (
1MISS 2
2MISS Earnings) to measure the extent to which earnings affect loan terms for firms with MBE=1 (firms with MBE=0).15 As such, if prospect theory can explain our findings in Tables 4 and 5, we predict that the MBE benefits functions are steepest around zero unexpected earnings benchmarks, and are concave in gains and convex in losses relative to earnings. Stated differently, we expect (1) that 2MBE(2MISS) is negative (positive), and (2) that 1MBE 22MBE Earnings <1MISS 22MISS Earnings. Table 7 summaries the results for models (3a) and (3b).The left side of Table 7 describes the results of the full sample firms. For our spread analyses, Panel A indicates that the coefficient of Earnings2, 2MBE(2MISS), is negative (positive) among firms meeting (missing) expectations, as predicted. To test our second prediction, we first calculate the estimated value of
15 The reason for doing so is that the slope of a quadratic function ax2+bx+c is 2ax+b.
26 Earnings
MBE
MBE 2
1 2
for every observation with MBE=1 from equation (3a).
Similarly, we calculate the estimated value of 1MISS 22MISSEarningsfor every observation with MBE=0 from equation (3b). Next, we compare the mean (median) of estimated values of 1MBE 22MBEEarnings and that of
are significantly larger than those of
Earnings
MISS
MISS 2
1 2
, contrary to our predictions.
As mentioned above, the slope of the value function in prospect theory is steepest near the reference point. Therefore, the inconsistent results above may be driven by the inclusion of firms reporting earnings by large amounts. As such, we restrict our sample into firm-quarters with earnings surprises that do not exceed the absolute value of 0.005 (Burgstahler and Dichev 1997).16
The right-hand side of Table 7 shows the results of this subsample. It can be seen that the coefficient of Earnings2, 2MBE(2MISS ), is significantly negative (positive) among firms meeting (missing) expectations, as predicted. In addition, the mean (median) of 1MBE 22MBEEarnings are significantly smaller than that of
Earnings
MISS
MISS 2
1 2
, in line with our predictions. More specifically,
untabulated results show that about 99% of the estimated values of Earnings
MBE
MBE 2
1 2
and 1MISS 22MISS Earningsare positive.
Overall, our empirical results provide direct evidence that even though banks are unlikely to use earnings as a heuristic or reference point to determine loan spread
16 We also use the scaled earnings surprise with internal widths of 0.0025, and the results remain qualitatively unchanged.
27
due to superior access to information about borrowers, the prospect theory framework still can explain their loan behaviors. The findings in this section are in line with the argument by Degeorge et al. (1999) that there is a salient dividing line between meeting and missing the norm and saliency makes the norm itself a focal point, which reinforces lenders’ psychological properties.
[INSERT Table 7 HERE]
The effects of other earnings thresholds on loan terms
Brown and Caylor (2005) show that, in recent years, managers’ propensity to meet analysts’ earnings expectations exceeds their propensity to avoid either earnings losses or earnings decreases. Thus, to shed light on whether the findings on MBE in public financial markets can generalize to private loan market, we focus our analyses on the analysts’ earnings forecasts benchmark. In this section, we further explore the effects of other two earnings thresholds on loan terms--- avoidance of losses and earnings decreases.
To test for the effect of avoidance of earnings losses on loan terms, we replace
MBE with AvoidLossess in equations (1) and (2). AvoidLossess is an indicator variable
taking the value of one if a firm’s basic earnings per share before extraordinary items is greater than or equal to 0 at quarter q in year t, and zero otherwise. In addition, for the set of Borrower_Specific control variables in equations (1) and (2), we replace
EarSurp with EPS to control for the effect of earnings on loan terms. EPS is defined
to be a firm’s basic earnings per share before extraordinary items at quarter q in year t divided by its stock price at the beginning of quarter q. All other variables are as previously defined in equations (1) and (2).
Similarly, to examine the effect of avoidance of earnings decreases, we replace
28
MBE with AvoidDecreases in equations (1) and (2). AvoidDecreases is an indicator
variable taking the value of one if a firm’s earnings per share before extraordinary items at quarter q in year t is greater than or equal to the same in year t-1, and zero otherwise. Further, we replace EarSurp with EarChanges to control for the effect of earnings on loan terms, where EarChanges is defined to the change in a firm’s basic earnings per share before extraordinary items between quarter q in year t and quarter q in year t-1, divided by its stock price at the beginning of quarter q in year t-1. All other variables are as previously defined.
The results are presented in Table 8. For brevity, we do not report the empirical results of control variables. Panels A indicates that irrespective of avoidance of earnings losses or earnings decreases, meeting or beating earnings thresholds has a lower spread and longer maturity. Panel B shows that all results are broadly comparable to those in Table 5, indicating that regardless of reporting profits or reporting earnings increases, the benefits of meeting thresholds are more pronounced for firms in financial distress than for financially sound firms.
Overall, the findings in Table 8 indicate that although reporting profits and reporting earnings increases are associated with favorable loan terms, their effects are stronger when firms are in financial distress.
[INSERT TABLE 8 HERE]