JaNAURY 15 2003 HG604—session 2
Handouts:
1. Livingston paper 3:30-5, CG 2063. Your memo: Due in class next Wed. 900-1100 words. How many Aphorisms suggestiosn does he ignore?
2. Martin Chapters 6-7.
THE BAIN REGRESSION:
Profitability = alpha + beta*concentration
What problems are there? Askt he class, and then go on to talk about them.
MARTIN: BROZELL PROBLEMS:
1. Disequilibrium? No.
2. Bias in industry selection. No.
3. Firm selection bias. INdustries with small firms would not be included. That’s OK.
BIG PROBLEM 1: Serial correlation. This is not mentioned in Martin.
Some industries have high profits at the same time because of omitted variables that are correlated. So the data sample is really smaller than it seems. Some observations are:
Motor vehicles
Washing machines
Steel works and rolling mills
Cast iron pipe
Wire
Doors and shutters, metal
Are these independent disturbances? No. The price if iron is in all of them.
BIG PROBLEM 2: UNIT OF ANALYSIS: (not in Martin)
Bain used industries such as Cigarettes, Soap, Paper Goods. He used government definitions, throwing out some clearly wrong ones (Cane sugar vs. Beet sugar) since demand, not supply, is what is relevant here.
He averaged together the profitabilities.
Firms could be used instead. Which is better? Firms--- more info. Use a separate indsustry dummy.
BIG PROBLEM 3: LEVERAGE: (not in Martin)
This is an omitted variable problem. More risky industries will need higher returns.
There are two ways that a firm can get capital, by DEBT and by EQUITY. Equity is what the owners put in, and debt is what they borrow. Let us use an example to illustrate what this means.
Some people form a corporation by putting in 1000 dollars with which to buy capital. They use it to buy sewing machines. The corporation
has 1000 shares, with an initial value of 1 dollar each. Each shareholder gets as many shares as he put dollars into the company. Each share has one vote for the choice of who will be on the board of directors that runs the company.
The company has no debt, so we say it is UNLEVERAGED.
If the company has net revenue ("net" meaning after variable costs) of $200 in the first year, the return on equity is +20%. The return
on assets is the same, since the company has no debt.
If the company's net revenue had been $50, the return on equity would have been 5%.
The book value of the equity is 1000 dollars, and so is the market value, at the start of the firm. Suppose the price of sewing machines falls in half. The company's assets now have a market value of only 500 dollars, so the stock price will fall to 50 cents per share, and the market value falls to 500 dollars. The book value of equity is still 1000 dollars, however.
If the shareholders want to, they can revise the book value. They do this by "writing down" the assets by $500. But they do not have to do that, and companies only write down assets occasionally.
Now let us introduce debt. Suppose the price of sewing machines goes back up, so the assets are worth $1000 again. The directors decide to borrow $2000 from a bank, at an interest rate of 15%. The company is now "highly leveraged".
The company has revenues of $600 the next year, because it has tripled in size and the return on assets is still 20%. The company must pay $300 in interest to the bank, though, which leaves $300 in cash flow for the shareholders. Thus, the return on equity is 30%.
Suppose the net revenue had been $150 (a 5% return on assets). The company must pay $300 in interest to the bank, which leaves -$150 for the shareholders (the company would have to sell some sewing machines to come up with the money). The return on equity would be - 15%.
This example illustrates how leverage increases the riskiness of the company's stock even though it does not increase the riskiness of the company's assets. An unleveraged company would have had a return on equity of either 20% or 5%. The leveraged company has a return of either 30% or -15%.
Thus, any company can affect the riskiness of its stock by deciding how much debt to hold.
The "residual claimants" of a company are the people who get whatever profits are left over once all the debts are paid. In this example, they are the shareholders. The bank has first claim on the cash flow, and the shareholders are legally allowed to keep only money in excess of the interest payments. The residual claimants have the riskiest claims. The bank still runs some risk---it could be that the company loses $1100 in one year, for example, so it cannot pay the $300 in interest even if it sells off assets-- but the bank's risk is less than the shareholders'.
Note, too, however, the following:
1. The downside risk of the shareholders is limited to losing the $1000 that they invested in the company.
2. The downside risk of the bank is limited to losing the $2000 loan it made.
3. The upside gain of the shareholder is unlimited. If the company earns $10,000, then after paying the bank $300 in interest, the shareholders keep all the excess.
4. The upside gain of the bank is limited to the $300 interest it was promised.
BIG PROBLEM 4: DEMSETZ CRITIQUE (Simultaneity)
Demsetz Critique. Suppose some firms have low costs. They will grow, and the market will become concentrated.
(Simultaneity. Concentration might depend on profitability. Simultaneity is another standard problem in regression analysis. )
To test this, do a regression at the firm level:
Profitability = alpha + beta*concentration + gamma*market_share + industry_dummy
If you run this, it turns out that beta is insignificant.
Does it matter that market share is not independnet between obeservations? No.
The problem Demsetz really points to is lack of theory (another general problem). Why are some industries more concentrated? Why are some industries more collusive?
I would like to use the following categories of technology:
A. Constant marginal cost, identical across firms
B. Constant marginal cost, different across firms
C. Rising MC, different across firms, in an Elbow Curve (my neologism)
I would like to use the following categories of behavior:
1. Bertrand. A Nash equilibrium in price.
2. Cournot. A Nash equilibrium in quantity.
3. Collusion without side payments. A cooperative equilibrium.
Let’s suppose, as Martin does, that there are just two potential firms, one with high cost and one with low cost (if they aren’t identical)
BAIN (in my interpretation): A. Constant marginal cost, identical across firms
We can’t be sure this is what Bain meant, and Martin gives quotations from Bain to indicate otherwise, but since Bain had no model, he was free to say contradictory things, and not really know what he meant himself. Keynes was the same way; thus, the silly arguments over what Keynes meant, and the famous Hicks IS-LM model trying to interpret Keynes.
A1. P=c . 2 firms. Zero profits.
A2. P>c, 2 firms. Positive profits.
A3. P >c, 2 firms. Biggest possible profits.
Then, the regression tests for A2 and A3 versus A1.
MARTIN. B. Constant marginal cost, different across firms
B1. P=c_h . 1 firm. positive profits.
B2. P>c_h, 2 firms. Positive profits. Firm L has higher profits.
B3. P >c_h, 2 firms. Biggest possible profits. Firm H has higher profits.
Then, we expect, roughly, both market share and concentration to be significant. Market share should be the most important.
RASMUSEN. . C. Rising MC, different across firms, in an Elbow Curve (my neologism)
C1. Pc_h, randomized . 2 firms . positive expected profits. This model has not been adequately analyzed. 1922 Edgeworth Paradox, item 62 in the Reader.
C2. P>c_h, 2 firms. Positive profits. Firm L has higher profits.
C3. P >c_h, 2 firms. Biggest possible profits. Firm H has higher profits.
This model has 2 firms always active. Then, we expect, roughly, both market share and concentration to be significant. Market share should be the most important.
BIG PROBLEM 5: ACCOUNTING
Profit rates vary across industries because of the accounting rules and the types of expenses. The problem arises because costs and revenues arrive at different times. Suppose two firms each have 100 in capital.
Firm 1 pays 50 for labor and raw materials and gets 80 in revenue each year. Profit is 30, and the return on capital is 30%.
Over two years, total profit is 60.
Firm 2 pays 50 for labor and 60 for raw materials inventory in the first year, and gets revenue of 110. Profit is 0 and the return on capital is 0%.
Firm 2 pays 50 for labor and 0 for raw materials in the second year, and gets revenue of 110. Profit is 60 and the return on capital is 60%.
Over two years, total profit is 60.
Growing industries will look less profitable.
How this plays out depends on the particular accounting rules. But what is general is that differetn industries will be affected differently.
At the end:
COnjectural variation
Lerner Index