Does Leverage Create Price Stickiness?

SRW has argued that leveraged firms are less likely to lower prices as they need to make debt payments, and this can cause price stickiness.

I think this is a rabbit hole.

First, prices are not sticky — they just don’t respond to macro shocks in the way that simple models predict. That is a problem with the model, not with prices.

Second, whether a firm finances itself with debt or with equity, it must still meet its overall cost of capital. Firms are not more free to cut prices if they are financed with debt versus equity. In both cases, firms have committed in advance to deliver a certain operating margin, and if they are not able to meet this commitment, the management of the firm is replaced and/or the cost of financing for the firm is increased, up until the new, smaller firm *is* able to meet the margin demanded.

That is true regardless of how the firm is financed. The only difference is the mechanisms applied — bankruptcy + replacement of management versus replacement of management.

Let’s look at the data. I scraped Yahoo finance for the financials of the top 1200 firms for which there was a (north america) NAICS code.

Of that population, about 900 firms had a current 5 digit NAICS industry code producer price time-series tracked by the BLS on a monthly basis since at least 2007.  Recall that producer prices are not the prices paid by producers, but the prices received byproducers for their output. The BLS receives over 100,000 price observations a month and tracks price indexes for up to 6 digit NAICS industry codes, which are rolled up into 5 digit codes, etc, finally being rolled up into the headline producer price index.

Of that population, about 800 were non-financial firms (e.g. everything except 52*).

Then, I compared the average firm financials for the trailing 3 year period with the standard deviation of the percent change in price index for the firm’s 5 digit NAICS industry code.

Firm financials were smoothed, i.e. the average leverage was the sum of the assets over the 3 year period divided by the sum of the equity over the 3 year period, to mitigate issues with  equity volatility. The average long term debt to equity values were calculated the same way.

The results are zero correlation between the debt burden or leverage of a firm and the standard deviation of price changes for the firm’s products:

Here is the same data but with book value used in place of equity when calculating debt burdens and leverage burdens. Again, there is no relationship.

There is zero evidence for the leverage-price stickiness theory.

Capital is not free. Firms primarily financed with equity do not have more freedom to lower their markups than firms financed with debt.

UPDATED:

Added the debt to book and assets to book graphs as well.

About these ads

12 responses to “Does Leverage Create Price Stickiness?

  1. Pingback: Standard Deviation | web-trends.net

  2. RSJ,

    Wow, thanks. I’m very grateful that you’ve taken so much time and trouble to look into this.

    Almost apologetically, though, I’ll nitpick and say that you’ve not convinced me that I’m wrong.

    On the theoretical point, sure, equity demands a return, just as debt does. But the institutional differences between debt and equity are consequential. Equity can make its demand only ex ante. After an unanticipated shock (and putting aside agency issues), returns to equityholders are worked out to maximize the forward-looking value of equity, which requires downward price adjustment. A profit-maximizing all-equity firm does drop prices in order to maximize forward-looking value. The same firm when leveraged drops prices less, under some assumptions about how the distribution of sales varies with price. This is trivial to demonstrate — see for example a worked example in my response to Scott Sumner. It’s skeletal, but it carries the intuition.

    Even in theory, managers incentives are equity-like rather than debt-like in the face of a systemic or industry-wide demand shock. That is, a manager who fails to deliver on shareholders ex ante return expectations may be fired if her competitors have performed as expected. But rational shareholders would not fire a manager who underperforms original expectations in the context of industry or global underperformance. The population of managers as a whole has underperformed; shareholders have no evidence that failure to meet expectations was due to any inadequacy of their staff unless their firm underperforms the industry. But the strategy I propose is the value maximizing strategy in the context of an industry-wide demand shock.

    In practice, it is unwise to treat the fact of ex ante return expectations as evidence that managers face similar incentives with regard to shareholders as they do with regard to creditors even when firms do underperform their industry. Agency costs and attenuation of control are real phenomena, managers can underperform and remain entrenched for significant periods. Share prices adjust to disappointments by falling to harmonize return expectations before organizations adjust via management changes (that might restore lost value). This is not to say that shareholders or acquirers do not, eventually, hold underperforming managers accountable. But it is to say that the constraint is much weaker than the constraints imposed by creditors who are able to enforce payment via bankruptcy.

    Equity capital most certainly is not free. In expectation looking forward and averaged over time, it is more expensive than debt capital. But equity-financed firms very much do have more freedom to modulate mark-ups in response to changes in market conditions than predominantly debt-financed firms.

    Re the empirical part, I’d like to be persuaded, because you’ve done a great deal of work. And I’ll certainly concede that all I’ve done is offered a conjecture, and what I believe to be a plausible theory about why the conjecture might be true. But my conjecture is just that: I certainly haven’t demonstrated that it isn’t a “rabbit hole”, as you say.

    But I am not dissuaded from my conjecture, despite your experiment. Your work is above and beyond the call, as a response to a blog post. But two basic flaws in its design leave me stuck on my theoretical priors:

    1) You are measuring (rough) price changes on an industry basis, but leverage on a firm basis. As I pointed out in the original piece, it is industry-wide leverage that matters. If leverage among firms offering close substitutes is widely dispersed, then a highly levered firm that tries to hold its price sees expected sales fall to near zero.

    Let’s unpack that. As firms decide whether and how to adjust prices following a demand shock, they try to maximize expected profit that can be appropriated by shareholders at different prices. Payoffs to creditors are excluded from consideration. Payoffs appropriable by shareholders at a given price (above marginal cost) are weakly increasing in the expected value of sales, but also depend upon the variance and skewness of sales. (Weakly, because there is a region of indifference: payoffs below a certain threshold are worth zero to shareholders.) We face a demand shock after which an unlevered firm would increase both sales and profits by dropping prices. But since sales are stochastic, that price drop might well reduce expected profit appropriable by shareholders of a levered firm, by drawing in the right tail of the EBIT distribution. The probability mass of the right tail is what shareholders of a leveraged firm are maximizing, full stop.

    (I’m using “right tail” to mean the portion of the distribution of EBIT outside the bankruptcy regime; whether that actually looks like a tail or includes the hump depends on how near to bankruptcy our leveraged firm is.)

    In an industry with both levered and unlevered firms, I claim the levered firms will go bust with near certainty following a large negative demand shock. Holding prices firm doesn’t work, because competitors drop prices, so both the expected value of our sales and the mass in our above-bankruptcy tail fall precipitously.

    Basically, with widely dispersed leverage, more highly indebted firms face what to a first approximation is Bertrand competition. If they fail to match their less leveraged competitors’ price drops, they lose nearly all sales and go bust. If they do match less leveraged competitors, they still go bust, as they are insufficiently profitable at those prices. So, one way or another, they go bust. They face a choice between a rock and a hard place, a matter of staking their existence on one of two different lottery tickets that both have terrible odds. (One choice may still be better than another, depending inter alia on whether the firm is capacity constrained.)

    But if leverage within an industry is not widely dispersed and the number of large players is moderate, firms will compete strategically. Leverage will serve as an implicit coordination device: each firm will make pricing decisions taking into account the fact that their competitors face precisely the same dilemma. Ultimately my claim (and this would be fun to work up the math for) is that a Nash equilibrium exists among similarly leveraged competitors under which all firms refuse to drop prices below a threshhold that ensures mutual bankruptcy. This equilibrium won’t exist if there is even one firm with substantial spare capacity that is unlevered — that firm will win by dropping prices and capturing the market. But if nearly all the firms are leveraged, or if the production capacity of the less-levered tail is small relative to the size of the industry, firms will find a minimum price that permits some winners to survive, and implicitly coordinate at that price.

    (Even if there are a few less levered firms, they may join troubled competitors in holding a higher-than-profit-maximizing price for fear that bankrupted rivals would reorganize with lower cost structures, turning today’s advantage into tomorrow’s trouble. But let’s keep things simple.)

    So it is not enough to look at firm leverage individually. You want to look at industries, industry leverage, and dispersion of industry leverage, perhaps weighting by market share. My claim is that industries that are highly levered and in which there is little leverage dispersion will exhibit downward price stickiness. I make no claim about any individual firm unless it is in a tightly leveraged industry.

    Before we go to the next point, I want to separate concerns a bit. For any individual firm, holding the shape of the distribution of sales constant, leverage reduces the degree to which the firm will reduce prices in the face of a negative demand shock.

    However, the shape of the distribution of sales is not constant with respect to changes in price. My claim is that industry structure determines the degree to which the shape of the distribution of sales will be adversely affected should a firm refuse to drop prices. (“Adversely affected” means a diminishment of EBIT mass in the right tail.) In concentrated, leveraged industries I argue that the adverse effect of refusal will be small. In industries with firms of heterogeneous leverage, the adverse effect of refusal will be large, and we won’t see a relationship between leverage and price stickiness.

    2) You are measuring price dispersion symmetrically. The logic that compels a firm to reduce prices less than an unleveraged firm does not make clear predictions about the effect of a positive demand shock. There is little reason to think that highly leveraged firms in highly leveraged industries will fail to raise prices as readily as less levered firms. Even in response to a positive demand shock, a leveraged firm might behave differently than an unleveraged firm, as it still faces a different maximization problem than an unleveraged firm would. But positive demand shocks bring firms-as-call-options deeper “into the money”, where they tend to behave more like the underlying (in this case the assets of the firm as an unleveraged entity). Further, following a positive demand shock, it strikes me as hard, even in industries whose members have quite similar capital structures, to tacitly coordinate. I have reason to assume that my competitor won’t drop her price to a level that would likely ensure her bankruptcy. But the same competitor might reasonably under-adjust to a positive demand shock (in hopes of capturing ongoing market share) or pick a price higher than an unleveraged firm would choose (if the positive demand shock is associated with a wider expected distribution of sales at high prices).

    Anyway, we need an asymmetric measure of price rigidity. The only strong claim is sticky-downward. (Lubricated upward is not impossible.)

    So, if we wanted to run an experiment, I think we should…

    1) look at industries full-stop, and prices within industries, rather than individual firms;

    2) include measures of leverage dispersion and industry concentration in our model;

    3) interact average leverage with those two variables.

    4) use an asymmetric measure as our dependent variable (We might consider semivariances, even semikurtoses, rather than straight variance. We might try probits of the probability of unusually large downward spikes, as we’d expect even leveraged industries to maintain margins of safety sufficient for moderate adjustments, so there might be a sharply kinked effect.)

    I think we’d expect no effect of leverage not-interacted, no effect of leverage dispersion not-interacted, a diminishment of price adjustment associated with uninteracted concentration (for reasons unrelated to leverage, simple pricing power), an effect of the interaction of leverage on both dispersion and concentration, and especially on the three way interaction.

    We probably still disagree about the merits of my conjecture. But regardless, I really appreciate your putting so much time and effort into considering and testing it. I’m a big fan of your work in general, in comments everywhere and here at windyanabasis.

    • Thanks for the comment!

      There is a lot to unpack there, and it will take me some to do it. However, ask yourself why do shareholders sell equity in anticipation of a broad decline in earnings? If it would truly be irrational for them to punish individual managers as a result of a market downturn in profitability, then they would not sell stock.

      But they re-price the stock up until the the expected return is the return demanded, rather than lowering their return demands. As a result, the market value of every firm shrinks, and this translates into a physical shrinking of the capital stock as well, at least on the margin. That is it leads to a reduction in investment, which means less output and employment. You don’t need sticky prices for that, and you don’t need differences in capital structure either.

  3. Pingback: interfluidity » Leverage and sticky prices — am I wrong?

  4. What’s going on at 3.5%?

  5. good question about 3.5%.

    • As SRW pointed out, I am measuring leverage on a firm basis but price deviations on an industry basis. When there are several firms in the same industry (each with slightly different leverage), you will get several (close together) dots with the same vertical position.

      In the next post (in a few minutes) I’ll consolidate leverage across industries as well. Stay tuned.

  6. Of course the NAICS groups with the highest gross margins are where you’re likely to find monopolies (Lerner monopoly index and all that).

    http://en.wikipedia.org/wiki/Lerner_index

    Hmm, there should be a tax for that. :o)

    http://traderscrucible.com/2011/04/21/yglesias-and-mmt/#comment-350

    • In general, when it comes listed companies that have access to the debt markets, it is rare to find more than a handful of firms at the 5 digit industry level. But I think that makes the data better, not worse, because it means that the price index is more likely to measure the prices received by the actual firm. Ideally you would have one price index per firm, but one price index for a small group of firms is the next best thing.

      • The point of tackling monopoly pricing (to connect the dots for other readers) is to reduce cost-push inflation with an iteration of what Stanley Weintraub dubbed “tax-based incomes policy” (TIP)– though Weintraub targeted wage hikes instead of value added pricing (it was the 70s, people were craazzyy). :o)

  7. Pingback: interfluidity » Visualizing Keynesian & Monetarist recessions

  8. Pingback: Visualizing Keynesian & Monetarist recessions : Invest My Money

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s