Positive valuation shocks predict bond issuance but not stock issuance
The disciplinary force of arbitrage capital?
When you buy or sell a stock no money reaches the firm. When its stock price goes up, the firm feels richer but no new capital is immediately available for real investment. All that happens is that the firm’s mark-to-market valuation goes up so that it may, possibly, market conditions permitting, raise more capital. In order to do so, the firm would have to issue new stock or bonds.
Since the net worth of the firm has increased but its debt obligations are the same on paper, its market leverage has fallen. In other words, it is now under-leveraged. Maximizing shareholder value dictates that it issue more debt — otherwise it would be playing too safe for its target credit rating/probability of default. (And, no, the Modigliani–Miller theorem does not hold in the real world.) It could also disobey the neoliberal dictum and issue stock. But that would dilute the net worth of its shareholders, exposing it to predatory buyout firms and activist hedge funds insistent on unlocking shareholder value. In this manner, arbitrage capital, ie heavily-capitalized firms that specialize in holding dry powder to buy up illiquid assets, has exercised discipline over industrial firms since at least the early-1980s.
This is what the logic of discipline suggests. But what do firms do in actual practice? Here we test the main prediction of the logic outlined above. Specifically, we ask whether US nonfinancial firms respond to positive valuation shocks by issuing debt or equity. We show that positive valuation shocks predict debt issuance but not equity issuance. We document this result for a variety of valuation metrics and with and without controlling for CEO confidence. The latter serves as a control since high executive confidence in the firm’s future earnings prospects should predict security issuance independent of the market valuation of the firm. We show that our result is robust to controlling for CEO confidence. Our result suggests that the disciplinary force of the market for corporate control introduces a strong bias towards ever-greater leverage in a macro environment where market-wide risk premia get ever-compressed and valuations seemingly keep going up.
We obtain quarterly data from 1987Q2 to 2020Q4 for debt and equity issuance by US nonfinancial firms, Conference Board’s CEO confidence index, the market capitalization of SP500 firms, the cyclically-adjusted price-earnings ratio, and the Shiller ratio from Haver Analytics. Figure 1 displays the raw series. Note the extreme rise in security issuance in 2020 — firms really gorged on external finance as the cost of capital plummeted, perhaps due to unprecedented liquidity provision by the Fed.
In order to achieve stationarity, we first-difference CEO confidence, and we log-transform and first-difference all the other time-series. Figure 2 displays the detrended series. Visual inspection shows that the detrended series are all stationary.
Our hypothesis is that positive shocks to market valuation predict debt issuance but not equity issuance. In order to test our hypothesis, we run a number of vector autoregressions and carry out Granger causality tests. The null of the Granger test is that the lagged values of one series do not predict the other after controlling for the latter’s own lagged values. If we can reject the null with high confidence, we can be sure that one predicts the other. Table 1 displays the results.
We find that all valuation metrics uniformly predict bond issuance but not stock issuance. We can reject the null at the 1 percent level for bond issuance no matter which proxy we use and whether or not we control for CEO confidence (0.001 < P < 0.009). In sharp contrast, no matter which valuation metric we use and whether or not we control for CEO confidence, we cannot reject the null for equity issuance even at the 10 percent level (0.450 < P < 0.989). The strongest predictor of bond issuance is the cyclically-adjusted price-earnings ratio for the SP500 (F-stat = 10.37, P = 0.001). This remains the case after controlling for CEO confidence (F-stat = 8.07, P = 0.005). So there is very strong empirical support for our main result.
The finding that the CAPE contains the strongest predictive information on 1-qtr forward bond issuance may be useful in pricing corporate bonds. Our main finding, that nonfinancial firms respond to positive valuation shocks by issuing more debt is highly significant. Given the secular decline in the cost of capital (see the following graph, stolen from here), and the secular rise in valuations, especially since the Western financial crisis, it is clear why firms have levered up to unprecedented levels (next figure).