Monday, June 30, 2014

Risk & Valuations, Part 3: Low interest rates do NOT lead to higher leverage (in practice)

My previous post described a model that would call for more debt in a corporate capital structure as debt becomes more expensive.  So, how does this play out in the real world?

It's a little tricky to pluck out of the data because there are cyclical factors, supply factors, monetary policy factors, etc., and all these factors are endlessly tangled up with interest rates and equity premiums.  But, I think I can at least suggest plausibility.

As a reminder, here is my graph from the previous post, demonstrating the counterintuitive consequences of changing debt and equity rates.  This was based on valuing a firm with the Black-Scholes model, treating the entire firm as the underlying asset, the equity as the call option, and the interest expense of the debt as the premium on the option.

This model predicts:

Falling Unlevered Equity Risk Premium (UERP) leads to:
(1) steeply increasing Enterprise Value (EV) (proxied in the graph by Share Price)
(2) slightly increasing leverage.

Falling Risk Free Interest Rates (RFR) lead to:
(1) increasing EV, though not as steeply as UERP
(2) steeply decreasing leverage.

If UERP declines and RFR increases by an equal amount, the net effect is a sharp increase in leverage and a small increase in EV.

In order to test the idea on historical data, I used data from the Federal Reserve Z.1 Financial Accounts report.  Using annual data beginning in 1960, from the Nonfinancial Corporate tables, I used profit before and after tax, interest paid, corporate equity, and credit market instruments from the Financial Accounts report, and added the 10 year Treasury Yield, the GDP Price Deflator, and  analyst growth estimates from NYU's Aswath Damodaran, who maintains several important data sets.  I use 10 year treasury rates because non-financial corporate interest expenses appear to track this rate.

I was able to derive several descriptive variables of nonfinancial US corporations over time, including ERP, UERP, and implied PE ratios.  The rates for corporate debt and the ERPs seem to be much higher for this data than they would be for, say, the S&P500, which is understandable.  But, the patterns appear to be similar.  Here is a comparison of my derived UERP to Damodaran's ERP, which I have adjusted to account for leverage.  Though my z.1 derived UERP is higher, the patterns are similar.

Below is a review of corporate leverage and capital premiums over the past 52 years.  The patterns roughly fit the counterintuitive predictions of the model.  (Enterprise Value is the combined value of debt and equity.)

Keep in mind when comparing equity and debt levels, they are on a log scale.

Enterprise Value stagnates when UERP is high.  Leverage increases when RFR is high.  In the great moderation period, corporate leverage has been countercyclical.  Note that debt levels are fairly constant and that most of the change in leverage is from variations in equity value.  Note that when UERP declined in the 1980's while RFR remained relatively high, EV growth increased and leverage remained high.  It was only after RFR continued to decline in the 1990's that debt declined.  Debt/Enterprise Value was at 47% in 1985 when RFR was 10.6% and it was still at 46% in 1990 when RFR was 8.6%.  (D/EV is shown in the graph below).  When RFR dropped to 6.4% by 1997, D/EV was down to 30%.  This was due to both a healthy increase in EV and a marked stagnation in debt growth.  After the shock to equity values in 2002, D/E quickly pulled back to 31% in the recovery, and, while debt started to increase in real terms after 2004, D/E continued to decline toward 30% as the economy grew.  Now, D/EV is again falling through the low 30%s as the economy stabilizes.

Both measures derived from Federal Reserve Z.1 report,
Non-financial Corporate Levels
For consistency in the peaks in EV and the duration of the stagnation, I compared 1968 to 2000 in the graph above.  In fact, leverage in 2000 was cyclically depressed due to high equity valuations stemming from growth expectations (see graph at right), so the cyclically neutral starting leverage of the low rate period was probably closer to 30%, and leverage is continuing to fall, and is now below 32%.

Note also that P/E ratios move inversely to leverage.  This is partly because when there are high growth expectations, equity values get bid up, as in the late 1990's.  But, it also reflects the fact that in low leverage contexts, equity has lower risk and volatility.  We tend to think of speculative firms as high PE firms, because the high PE is driven by growth expectations (like in the late 1990's).  But, imagine a firm that issued shares and simply invested them in short term treasuries, with a 100% payout ratio.  If it earned 2% returns, it would trade at a PE of 50.  So, high PE ratios during times of low interest rates aren't the result of firms fattening up on cheap debt.  Rather, they are a result of low leverage and a lower equity premium for the average firm.

Look again at the graph above of the Unlevered Equity Risk Premium (UERP) over time.  Notice how it was relatively high in the late 1970's and is relatively high now, and was low in the 1980's & 1990's.  Now, look at the smaller chart on PE Ratios and Debt/Enterprise Value.  The late 1970's had PE ratios around 10, but lately, PE ratios have been in the teens.  This is partly because low interest rates cause values of all durable assets to rise (the discount rate in a CAPM model would be lower, for instance).  But, partly, this is because the lower interest rates counterintuitively lower leverage, which, in turn, lowers the required return on equities.  Note the same discrepancy in the 1980's and 1990's.  The UERP was very low throughout this 20 year period.  When interest rates were still high, leverage was still high, so PE ratios were held down.  But, when interest rates fell in the late 1990's, leverage fell dramatically.  The astronomical valuations at the time were partially a product of high growth rates, but even before factoring in the high expectations, PE ratios would have been extremely high.  If I adjust Damodaran's ERP for leverage, it is pretty stable from 1985 to 1997.  But, the market ERP, which reflects the market's typical leverage, fell by a full point during that time.  So, when risk free interest rates fall, using a CAPM-type valuation measure, there is a multiplier effect due to the fact that the Equity Premium might also fall with it, due to declining financial leverage.

So, when interest rates fall, we frequently see rising stock prices.  This is commonly attributed to firms boosting net earnings by leveraging cheap debt.  But, this simply does not bear out, empirically.  Lower rates do cause the value of productive assets to rise.  But, this is related to deleveraging.  And, thus, the rising stock market is related to lower risk.

When I divide the interest rate into an inflation premium and a real rate, the inflation premium is the stronger influence on corporate leverage.  This is because the real rate is a sort of price of debt that reflects both supply and demand.  The inflation premium carries the tax consequences of debt versus equity for a corporation more purely.  Note also that the two surges in D/EV since 2000, when interest rates were low, were from crashes in Enterprise Value, not from planned increases in Debt.

NOPAT = Net Operating Profit After Tax
10 Year Treasury Rate is on right scale, inverted
Here is one more graph, showing Equity and Debt over time, as a proportion of NOPAT.  NOPAT is Net Operating Profit After Tax.  It's profit after tax, but before interest expense - kind of the unleveraged profit of the firm.  What's interesting is that we think of interest rates as an incentive for debt utilization.  But, what we see here is that debt has been remarkably level through both extremes of interest rates.  The effect of lower interest rates on productive investment flows mostly through equity values.  And, maybe the tax issue is a red herring.  Maybe creditors are comfortable with a general relationship of Debt/NOPAT because avoiding default is paramount, and this keeps potential debt levels under a cap.  Equity is not encumbered by this concern, so that interest rate induced increases in enterprise values must be accommodated with equity.  Also, note that under about 5% (10 year treasury rates), interest rates seem to lose their power to increase equity values.  Maybe this is because rates in this range are suggestive of underlying economic problems or are associated with deflationary distortions of economic activity.  It does make one wonder how effective we could have expected Fed policy with the stated intention of pulling down long term interest rates well below 5% to be.

The ratios above tend to make cyclical effects hard to track, because the denominators tend to go batty during downturns.  So, here is Debt (Nonfinancial Corporate Credit Market Liabilities) as a percentage of Potential GDP, graphed alongside the Fed Funds Rate.  Note that the level of debt declines as the rate declines, remains low when the Fed Funds Rate bottoms, and only tends to rise again after the Fed Funds Rate has risen and plateaued.  Debt is rising now, as a proportion of Potential GDP, in spite of the low Fed Funds Rate, but as shown in the graph above, it's still declining as a percentage of corporate capital.  That's kind of the opposite of what everyone knows, isn't it?  Isn't there a story in the paper every day about how corporate profits are high because the Fed is enabling them to leverage up with cheap debt?

This reminds me of the idea I considered recently of Treasury Bonds and Real Estate as a sort of Giffen Good for savers.  I noted how levels of Real Estate and Securities in Bank Credit (government bonds) moved inversely to interest rates.  The levels of credit have been higher when rates have been lower.  Commercial & Industrial (C&I) loans as a proportion of GDP have not followed this pattern and have declined, just as Corporate Debt as a proportion of Enterprise Value has declined as interest rates have declined.  Could the divergence of corporate debt levels from these trends among other types of credit be related to this corporate tax issue that makes debt financing more desirable in high interest rate environments?

These peculiarities change the way we might imagine firms moving through the business cycle, which I will review next.

No comments:

Post a Comment