CH-16-17 Understanding Consumer Behavior – Macroeconomics : Chapter Notes

This Document Contains Chapters 16 to 17 CHAPTER 16 Understanding Consumer Behavior Notes to the Instructor Chapter Summary This chapter discusses the theory of the consumer. It is structured so as to follow, more or less, the history of thought on the topic. It first presents the Keynesian consumption function, then discusses Irving Fisher’s model of intertemporal choice, and then addresses the life-cycle and permanent-income theories of consumption. The chapter also includes a discussion of rationalexpectations theories of consumption. Comments This material probably requires two to three lectures. This is a good time to discuss the importance of microfoundations in macroeconomic models. I suggest to students that the material in Part V of the book allows us to both justify and to refine the important behavioral assumptions that underlie the earlier models. In the case of consumption theory, looking at the underlying microeconomics improves the performance of the short-run macroeconomic model. This chapter can be treated as an extended case study on the development of a theory in economics; it provides a good illustration of the productive interplay of theory and data. The primary motivation of the chapter is the puzzle of the difference between the short-run (crosssection) and long-run (time series) consumption function. The biggest difficulty in presenting this material is that the theory of intertemporal choice takes some time to develop; by the time it is complete, students may have forgotten the motivation. It is worth going back and reminding students of the puzzle at that point. Students also seem to dislike the “unrealistic” assumptions of the life-cycle model—in particular, the fact that, in a simple textbook treatment, we assume that people know their lifetime income and time of death with certainty (and we often assume a zero interest rate). I try, first, to reassure students that we do know how to relax these assumptions and, second, to point out that we have at least improved upon the naive Keynesian consumption function, which is even more unrealistic. I try to leave the students with the lesson that, in thinking about changes in income, two distinctions matter: temporary versus permanent and anticipated versus unanticipated. Use of the Web Site One important advantage of the Web-based software is that it allows students to see explicit lifecycle behavior with a nonzero interest rate and also to see how much difference interest rate changes can make. Students may also be surprised at how difficult it is to smooth their consumption over their lifetime. Students could also try to obtain something other than a completely smooth consumption path—one possibility would be consumption increasing gradually through the lifetime; another would be smooth consumption except for a blip somewhere in middle age (say, for financing children through college). 375 Use of the Dismal Scientist Web Site Go to the Dismal Scientist Web site and download quarterly data on the major components of personal consumption expenditures (durable, nondurable, and services) in the U.S. over the past ten years. Assess the effect of the recession of 2001 and the recession of 2008–2009 on consumer spending. Did it decline during these recessions? Did certain components decline but not others? The 2008–2009 recession was accompanied by a substantial decline in real household wealth. What components of consumption would you expect this to most affect? Do the data confirm your expectation? Chapter Supplements 16-1 The Components of Consumption 16-2 The Stock Market and Consumer Spending 16-3 Saving and the Fear of Nuclear War 16-4 The 1975 Tax Cut (Case Study) 16-5 Do Consumers Anticipate Changes in Social Security Benefits? (Case Study) 16-6 Is Unemployment Insurance Really an Automatic Stabilizer? 16-7 Additional Readings Lecture Notes Introduction Modern macroeconomics emphasizes the need for solid microeconomic foundations. Although we make many simplifications in macroeconomics and ignore most differences among individuals, we still hope that our macroeconomic theory is grounded in sensible microeconomics. Macroeconomics and microeconomics, after all, should not be completely separate disciplines; they are attempts to understand economic phenomena on different scales. This section of the textbook considers the microfoundations of consumption theory, investment theory, and money demand. These are all important ingredients of our model of the economy, and so we need a good understanding of them. A more detailed consideration of these elements of our models isolates the strengths and weaknesses of current theory and allows us to refine and improve our understanding of the economy. We start with consumption theory, since it is so central to our earlier analyses. The consumption–saving decision is the key behavioral element of the Solow growth model and also Supplement 16-1, underlies the multiplier effects in the short-run model. Consumption is also the biggest “The component of aggregate demand, accounting for about two-thirds of GDP. We cannot have a Components of Consumption” good understanding of the economy unless we have a good theory of consumption. 16-1 John Maynard Keynes and the Consumption Function The consumption function was introduced into macroeconomics by John Maynard Keynes in The General Theory of Employment, Interest and Money. We start our analysis by considering Keynes’s ideas about the consumption function. Keynes’s Conjectures Recall that our models are based on a simple consumption function: C = C(Y – T). (Y –T);C > 0; 0 < c 0 if MPK > r + δ. We can write this as a net investment function: In = In(MPK – (PK/P) (r + δ)). This says that net investment depends upon the profit rate, that is, the difference between the marginal product of capital and the cost of capital. The exact form of this function, and accordingly the response of investment, depends on the size of adjustment costs facing firms. Finally, to obtain total investment for the economy, we take the net investment function and add replacement of depreciated capital, which is simply δK. Our assumption that there are two types of firms is made for ease of exposition only. In the real world, where firms usually own rather than rent capital, the incentive to purchase new capital goods still depends upon the difference between the marginal product of capital and the cost of capital, just as our analysis suggests. We have now accomplished our most important task in the study of investment: We have explained why investment depends upon the interest rate. A higher interest rate makes it costlier Figure 17-3 for rental firms to own capital. Other things being equal, therefore, a higher interest rate reduces the profit rate per unit of capital, decreasing the incentive for rental firms to own capital. Investment depends negatively on the interest rate. Anything that increases the profit rate for any given rate of interest (such as a technological innovation that raises the marginal product of capital) shifts the demand for investment out. Supplement 17-1, Whenever net investment is positive, the capital stock is increasing and so the marginal “The Short Run product of capital is falling. This reduces the profit rate and decreases the incentive to invest. In and the Long the long run, increases in the capital stock thus drive the profit rate down to zero. When the Run: Investment and the Capital profit rate is zero, then net investment is also zero and the capital stock will be at its steady-state Stock” level. In this case MPK = (PK/P)(r + δ). The time it takes to adjust to the steady state depends on how quickly firms adjust their capital stocks, which itself is driven by the costs of building, delivering, and installing new capital. Taxes and Investment The incentive to invest is affected by provisions of the tax code. Corporate income taxes are taxes levied on corporate profits. If profits were measured, as our theory suggests, by the difference between the rental price of capital and the cost of capital, then the corporate income tax would not distort the investment decision. But profit, for tax purposes, is defined somewhat differently. Most important, the tax code permits firms to deduct depreciation only at historical cost, rather than at replacement cost. This means that depreciation is underestimated, and profit overestimated, in times of inflation. The government sometimes enacts policies to encourage investment. One such policy is the investment tax credit, which reduces a firm’s tax liability when it purchases capital goods. Case Study: Inversions and Corporate Tax Reform When an American company merges with a foreign one and reincorporates abroad, the merger is often referred to as a tax inversion. Although the reasons for mergers are many, an important one is to take advantage of favorable tax treatment by some other nations. While companies and their management should not be faulted for trying to increase their after-tax profits, the lost revenue to the U.S. Treasury means that everyone else either has to pay higher taxes or receive fewer government services. If tax inversions are indeed a problem, the fault should not be placed on the business leaders who are meeting their responsibilities to shareholders, but instead should be placed on the tax code, which provides incentives for these inversions. One tax reform that would reduce incentives for tax inversions is to lower the U.S. corporate tax rate to levels prevailing in other countries. The U.S. rate is about twice the average rate in Europe. Closing this gap would help reduce the incentive to incorporate abroad for tax purposes. Another reform is to tax U.S. corporations only on income earned in the United States rather than on income earned worldwide. Many countries, including Canada, France, Germany, Italy, and Japan have such territorial corporate tax systems rather than a worldwide one. Shifting the U.S. tax code to the international norm of territorial taxation might limit incentives for tax inversions. Such changes, however, would require raising other taxes to make up for lost revenue. Some economists have suggested shifting toward a tax on consumption, perhaps a value-added tax, which many European nations use. The Stock Market and Tobin’s q The economist James Tobin proposed a theory of investment that is distinct from, yet related to, the neoclassical theory of investment. He suggested that net investment depends upon a number, known as Tobin’s q: Market Value of Installed Capital q = Replacement Cost of Installed Capital Tobin’s q compares the value of the capital stock as reflected in the stock prices of firms with the value of the capital stock measured in terms of the current price of capital goods. Tobin argued that firms have an incentive to invest whenever this number exceeds one. If q > 1, then the value of capital, as measured by the stock market, exceeds the cost of that capital. Therefore, firms could increase their value by purchasing more capital. If the marginal product of capital is greater than the cost of capital, firms earn profits on the capital they own. Since these profits should ultimately influence the dividends paid out by firms, they raise the stock market value of firms. In this case, q will be relatively high. Thus, both the neoclassical theory and the q theory predict that investment will be positive when the profit rate is positive. Movements in the stock market are likely to be closely correlated with movements in q, since the replacement cost of capital changes only slowly. It follows that a rising stock market may be a signal of investors’ optimism about the economy and hence of the level of investment. This helps explain why economists and others pay so much attention to movements in the stock market. Figure 17-4 Case Study: The Stock Market as an Economic Indicator Supplement 17-2, As discussed in Chapter 10, the stock market is a leading indicator and so may help to predict “Asset Pricing I: the future course of the economy. Changes in stock prices certainly do not correspond perfectly Why Do We with changes in GDP, but the two are related. Care?” There are three reasons why stock prices and GDP fluctuate together:  Falling stock prices could indicate a decline in Tobin’s q as a result of investor pessimism about current or future profits. This change in investor sentiment would shift the investment function inward, reducing aggregate demand.  Falling stock prices reduce household wealth, resulting in a reduction in consumption and hence aggregate demand.  Falling stock prices could reflect bad news about technological progress and long-run growth, indicating that the natural rate of output (aggregate supply) will expand more slowly than previously expected. As a result of the link between stock prices and GDP, policymakers watch the stock market closely. Alternative Views of the Stock Market: The Efficient Markets Hypothesis Versus Keynes’s Beauty Contest Supplement 17-3, Economists continue to debate the reasons for movements in stock market prices. One view, “Asset Pricing II: known as the efficient markets hypothesis, assumes that a company’s stock price is a rational Stock Prices and valuation of a company’s value. According to this view, the stock market is informationally Efficient Markets” efficient, so changes in the stock price of a company reflect new information about the future Supplement 17-4, “Asset Pricing III: prospects for the company. Any information already known about a company is incorporated Bond Prices and into investors’ valuation of its stock, and so only “news” should affect stock prices. Supporters the Term of this view cite evidence that investors find it difficult to beat the market and that stock prices Structure of Interest Rates” tend to fluctuate randomly. Supplement 17-5, Another view questions whether the stock market is rational. Proponents of this view note “Asset Pricing IV: that much of the movement in stock prices cannot be easily attributed to news. They argue that Bubbles, Excess Volatility, and investors are less concerned about fundamental values of companies’ underlying stocks and Fads” more concerned about how other investors will behave. In other words, investors focus on what Supplement 17-6, other people believe about the valuation of a company rather than the company’s true worth. As “Asset Pricing V: a result, movements in stock prices are due to irrational shifts in sentiment, which Keynes The Capital-Asset Pricing Model” referred to as the “animal spirits” of investors. And since the stock market influences aggregate demand, these movements in stock prices contribute to short-run fluctuations in overall economic activity. Financing Constraints Supplement 17-7, Firms may finance their investment either through retained earnings (that is, profits not “Financing distributed as dividends) or by borrowing in financial markets. From the perspective of the Constraints in neoclassical theory of investment, these two are equivalent, since the theory assumes that firms Japanese Firms” face no difficulty in borrowing. In reality, however, firms sometimes are limited in their ability to borrow in financial markets; they face financing constraints. Such constraints are similar to the liquidity or borrowing constraints faced by consumers. If a firm faces financing constraints, then its ability to invest may be restricted when its profits fall, implying that investment may fall more during a recession. The role of financial intermediaries in providing financing for investment projects means that problems in the financial system often are linked with economic downturns. For example, when banks face difficulties, they may cut back on lending. Firms may then face a credit crunch—difficulty in borrowing to finance investment projects. In the IS–LM model, this is illustrated by a backward shift in the IS curve, resulting in a decline in aggregate demand. In the long run, a credit crunch may lead to misallocation of saving to its most productive uses, limiting the economy’s potential output of goods and services. Policymakers, therefore, keep a close watch on the soundness of the financial system. Chapter 20 of the textbook discusses in more detail the causes and effects of financial crises. 17-2 Residential Investment We now turn to the explanation of residential investment. The Stock Equilibrium and the Flow Supply At any time the stock of housing is fixed, so the supply of housing is fixed. The higher the Figure 17-5 relative price of housing (PH/P), the lower is the demand for housing. The intersection of supply and demand determines the relative price of housing. A higher relative price of housing, in turn, induces construction firms to build more houses. This model is related to our theory of business fixed investment. In particular, this theory resembles the q theory of investment. Just as the q theory suggests that the level of investment depends upon the market price of installed capital relative to the replacement cost, this theory suggests that the level of residential investment depends upon the market price of existing houses relative to the cost of building new houses, which depends upon the general price level. Figure 17-6 Also consider another extreme in which all housing is built and owned by landlords, who Supplement 17-8, “Taxes, Babies, would then be analogous to rental firms in our previous theory. (Many modern apartment and Housing” complexes essentially take this form.) The existing supply of housing and the demand for housing would determine the equilibrium rental price for housing; landlords and construction Figure 17-7 firms would have a greater incentive to supply more housing as the rental price increased. Just as in our previous theory, the incentive to supply new housing would depend on the relative price of housing as well as the real interest rate and the depreciation rate on housing. Supplement 17-9, Changes in Housing Demand “The Tax Treatment of If the demand for housing increases, perhaps because of population increases or economic Housing” prosperity, the relative price of housing rises. This increases residential investment. Decreases in the real interest rate also encourage residential investment, just as they encourage business fixed investment. When the interest rate falls, mortgage rates fall, which increases the demand for owner-occupied housing and encourages residential investment. Similarly, a fall in the interest rate reduces the cost of capital to landlords who build and own rental accommodations. Another key determinant of housing demand is credit availability. During the early to mid2000s, mortgage interest rates were low and mortgage loans were easy to obtain. Even households with questionable credit backgrounds were able to borrow— so-called subprime borrowers. House prices rose sharply, boosting residential investment. A few years later, many of these borrowers could not make their mortgage payments and so defaults on mortgages increased. This led to a tightening of credit and an increase in interest rates, lowering housing demand and house prices. The declining housing market pushed the economy into recession during 2008. 17-3 Inventory Investment The third main component of investment is inventory investment. Although it is small in Supplement 17-10, “The Importance of magnitude, it is of great interest to economists because it is so volatile and thus accounts for a Inventories” substantial portion of GDP fluctuation. Reasons for Holding Inventories Firms hold inventories for four main reasons. The first is production smoothing: Although the demand for a firm’s product may vary substantially over time, the firm might prefer to keep its production relatively constant. Manufacturers of snow blowers may find it more efficient to produce throughout the year and store snow blowers for sale in the winter, rather than attempt to produce all their output in the winter months. Inventories may also serve as a factor of production, in that they may increase the amount of output that a firm can produce. For example, inventories of spare parts allow manufacturing firms to avoid substantial downtime when machines break down. Third, firms carry inventories for stock-out avoidance; that is, they wish to have goods on hand to meet unexpectedly high demand. And, fourth, there is work-in-process or pipeline inventory, since partially completed goods in the production process are counted as inventory. Supplement 17-11, How the Real Interest Rate and Credit Conditions Affect Production Inventory Investment “Inventories and Smoothing” Supplement 17-12, Holding inventory is costly for firms. If an auto dealership holds a car on its lot for a month, it is “Production worse off than if it had sold the car, because it could then have placed the proceeds from the sale Smoothing and Coordination in the bank and earned interest on them. So inventory investment, like the other types of Failure” investment, depends negatively on the real interest rate. When real interest rates are high, firms Supplement 2-6, will try to hold less inventory. “Seasonal Adjustment and the Inventory investment also depends on credit availability. Firms typically rely on bank Seasonal Cycle” loans to pay for purchases of inventories. When loans are difficult to obtain, firms will reduce Supplement 17-13, the amount of inventory that they hold. With the onset of the financial crisis in 2008, credit “The Multiplier- Accelerator Model” became much tighter and firms reduced inventory investment sharply, allowing their inventory to actually decline during 2008 and 2009. As the financial system and economy began to recover, firms rebuilt inventory during 2010 and 2011. 17-4 Conclusion The models of this chapter reveal that all types of investment depend negatively on the real interest rate, thus justifying the simple investment function adopted earlier in the textbook. They also reveal that the level of investment depends upon other factors, such as the available technology and the government’s tax policies. Finally, they reveal that investment depends positively upon output growth. We thus obtain an interaction between the level of GDP and the growth rate of GDP: If aggregate demand grows, then investment will be higher, and higher investment in turn implies a higher level of aggregate demand. LECTURE SUPPLEMENT 17-1 The Short Run and the Long Run: Investment and the Capital Stock Saving, investment, and capital accumulation are discussed in a number of different places in the textbook. The classical model presented in Chapter 3 of the textbook explains how the real interest rate brings saving and investment into balance. The Solow growth model of Chapters 8 and 9 explains the long-run process of capital accumulation. The neoclassical model of investment presented in Chapter 17 explains the determinants of investment. How do all these models fit together? We start by reviewing the accounting issues. For simplicity, we suppose here that there is no inventory investment, and we do not worry about the distinction between business fixed investment and residential investment. Part of investment in any year goes to replace worn-out or depreciated capital. Thus total, or gross, investment (I) is the sum of net investment (In) and depreciation. If the depreciation rate is δ and K denotes the capital stock, then It = In,t + δKt. Also, the capital stock evolves over time according to the equation Kt+1 = (1 – δ)Kt + It = Kt + In,t. Net investment thus equals the change in the capital stock. The classical model assumes that the real rental price of capital is determined in the market for capital goods, implying that MPK = R/P. The classical model also notes that the real interest rate adjusts to bring about equilibrium in the loanablefunds market: S = I(r). The supply of saving is fixed over relatively short periods of time but depends in general on the level of output and thus upon the existing capital stock. The neoclassical model of investment, meanwhile, argues that net investment depends upon the difference between the rental price of capital and the cost of capital In = In (R/P – (r + δ)), where we have assumed that the relative price of capital equals 1 for convenience. If net investment is positive, then the capital stock is increasing. The economy is in steady state when the capital stock is not changing, or, in other words, when net investment equals zero. This occurs in turn when R/P = (r + δ). The Solow growth model teaches us that net investment equals zero when saving equals depreciation: sf (k) = δk. We can illustrate this in the loanable-funds market and the market for capital goods. Figure 1A shows the market for capital goods. In steady state, the rental price of capital equals the cost of capital, r + δ. Figure 1B shows the market for loans. In steady state, replacement investment exactly exhausts the supply of saving. Figures 2A and 2B illustrate the out-of-steady-state case where net investment is positive. Figure 2A shows that, at the current capital stock, the rental rate exceeds the cost of capital. Thus net investment is positive, as shown in Figure 2B. Positive net investment causes the supply of capital to increase over time, reducing the rental rate. The lower rental rate decreases the incentive to invest, so net investment falls (the In line shifts left). The higher capital stock implies that replacement investment increases. Also, the higher capital stock implies that GDP is higher, so the supply of saving increases, as shown in the right-hand diagram. ADVANCED TOPIC 17-2 Asset Pricing I: Why Do We Care? One important area of macroeconomic study is that of asset pricing, that is, trying to explain the equilibrium prices of various economic assets, such as stocks and bonds or houses. This involves the study of financial markets and turns out to present a number of problems and puzzles that have not yet been fully resolved. Yet it is an important area of macroeconomic inquiry because financial markets bring together savers and investors in the economy. Our hope is that financial markets do a good job of directing available loanable funds to those activities that are most profitable. Many economists do indeed believe that financial markets operate efficiently and are an important aid to the smooth functioning of the economy. Others are less sanguine and believe that irrational behavior in such markets may be a source of shocks that disrupt the economy. Robert Shiller, an economist who suspects that asset prices change in large measure because of capricious behavior of investors, explains the significance as follows : That prices change for no good reason is of great importance for many purposes. Prices of speculative assets guide very many economic activities in our society. When an asset is underpriced, incentives are created to neglect or abuse it. When it is overpriced, incentives are created to invest too much in it. The possibility that these prices may show repeated tendencies to move for no sensible reason matters greatly not only to those managing financial portfolios, but also to regulators, legislators, lawyers, corporate managers, builders, homeowners, collectors, conservators and others. A better understanding of the importance of such price changes may ultimately set the stage for people to take actions that will reduce their impact. Much work on asset pricing focuses on the stock market. Macroeconomists are particularly interested in the stock market for a number of reasons. First, movements in the stock market seem to be linked to movements in aggregate economic activity. Second, as noted previously, the stock market brings together savers and investors and thus helps guide the allocation of loanable funds to investment projects.2 Third, the stock market, if it functions efficiently, provides information about investors’ expectations concerning future economic performance. Macroeconomists are thus concerned by there being evidence of inefficiency in the stock market. There is one observation suggesting that lack of efficiency in the stock market actually might not be so serious. Most trading in the stock market is of existing shares, not new issues, and so has a less direct influence on the allocation of resources. Even if stock market prices do change for no good reason, these fluctuations in relative stock prices might then simply redistribute wealth from one set of gamblers to another, and the consequences for the macroeconomy might not be that large. Aggregate movements in the stock market still matter, however, because they represent changes in wealth, and wealth is a determinant of consumption behavior. ADVANCED TOPIC 17-3 Asset Pricing II: Stock Prices and Efficient Markets We discuss here some basic ideas about the pricing of a risky asset, focusing on the stock market. Consider a situation in which investors can invest in a riskless asset, which pays a certain (real) rate of return equal to r, or a stock, which pays an uncertain return. Ownership of a stock entitles the investor to a stream of dividends. Dollars received in the future, however, are worth less than dollars today, because dollars today can be invested at the interest rate r. An individual will be indifferent between a dollar today and 1/(1 + r) dollars tomorrow; this is the present discounted value (that is, the value today) of $1 tomorrow. Similarly, we can write the present discounted value of a stream of dividends as ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞3 ⎛ 1 ⎞4  ⎝1+ r⎟⎠ dt+1 +⎜⎝1+ r⎟⎠ dt+2 +⎜⎝1+ r⎠⎟ dt+3 +⎜⎝1+ r⎟⎠ dt+4 + PDV =⎜ where dt denotes the dividend at time t. Since the present discounted value tells us what the stream of dividends is worth today, we should expect it to equal the price of the stock today. To see how this works in more detail, suppose that ht is the holding return on the stock between period t and t + 1. The return on the stock consists of two components: the dividend that it will pay and any capital gain or loss that arises because of a change in its price. Thus, if pt is the price of the stock and dt+1 is the dividend, then ht = dt+1 + pt+1− pt pt pt Return = Dividend + Capital Gain. The return is uncertain because, at time t, investors do not know the price that the stock will command at t + 1 or the dividend that it will pay. For simplicity, however, let us suppose for the moment that investors have perfect foresight and so know the future dividend and the future price of the stock. In this case, arbitrage between the risky stock and the riskless asset will ensure that both earn the same return. This gives dt+1 + pt+1 − pt = r ht = pt pt ⇒ dt+1 + pt+1 − pt = rpt ⇒ pt (1+ r)= dt+1 + pt+1 = pt =⎜⎛ 1 ⎞ ⎝1+ r⎟⎠(dt+1 + pt+1). This is the basic equation for the pricing of a stock. It tells us that the price of a stock today depends upon the dividend it will pay next period and the price of the stock next period. An analogous equation will hold for the price of the stock next period:  1  +  1+r (dt+1+ pt+2). pt 1 = Substituting this into the previous equation, we can therefore write  1   1  pt = 1 +r dt+1+ 1 +r (dt+2 + pt+2). By repeated substitution of this kind, we can ultimately write the current price of the stock solely in terms of the future dividends2:  1   1  2  1  3  1  4 Pt =   1+r dt+1+ 1+r dt+2 + 1+r dt+3+ 1+r dt+4 +, The price of the stock is indeed just given by the present discounted value of the stream of dividends that the stock will pay. Investors don’t know future dividends with certainty, so the price of a stock actually depends upon their expectations of these dividends. If investors are risk neutral, meaning that they care only about average return and do not worry about risk, then arbitrage will imply that the expected holding return on the stock will equal the interest rate.3 The price of the stock is then the present discounted value of expected future dividends. If the market is efficient, then investors will be making the best possible forecast that they can of future dividends, given the information that is available to them. In this case, the price of the stock will reflect all the information available in the market about the likely profitability of the company.4 An important implication of this is that changes in stock prices should come about only as a result of new information. Thus, changes in stock prices should not be predictable on the basis of any currently available information, for if they were, arbitragers would be able to make profits.5 The overall value of the stock market, by extension, should reflect investors’ best predictions about the future profitability of all firms (and hence, among other things, about the future state of the U.S. economy). Economists who are skeptical about the efficiency of the stock market point to the stock market crashes of 1929 and 1987 as evidence of inefficiency. The Dow Jones index fell over 500 points in one day in October 1987. Although large fluctuations in stock prices are not themselves necessarily inconsistent with stock market efficiency, there was no obvious new information that could have led investors to revise their opinion about the probable future profitability of U.S. firms to such a dramatic extent (this drop represented a 22.6 percent fall in the market).6 Others have pointed out that information may be transmitted imperfectly within the market, so stock prices may change as a result of the transmission of news in the market. The market may rationally revise its view of the fundamentals even when no new outside news arrives.7 Chapter 5 of the textbook. 3 Supplement 17-6, “Asset Pricing V: The Capital-Asset Pricing Model,” considers how riskiness affects asset prices. 4 Different definitions of market efficiency are sometimes used depending upon exactly what information is reflected in the price of the stock. 5 This idea is sometimes loosely referred to as the random-walk theory: Stock prices should follow a random walk. See the discussion of rationalexpectations theories of consumption in Chapter 16 of the textbook for more discussion of random walks. See also S. LeRoy, “Efficient Capital Markets and Martingales,” Journal of Economic Literature 27 (December 1989) (particularly Section III) for a good discussion. 6 See Supplement 17-5, “Asset Pricing IV: Bubbles, Excess Volatility, and Fads,” for more discussion of stock market efficiency. 7 See, for example, D. Romer, “Rational Asset-Price Movement Without News, American Economic Review 83, no. 5 (December 1993): 1112–30. ADVANCED TOPIC 17-4 Asset Pricing III: Bond Prices and the Term Structure of Interest Rates The textbook speaks throughout of “the” interest rate. Yet we know that in the real world there are many different interest rates. An important first observation is that the simplifying assumption of a single interest rate is not badly misleading for much macroeconomic analysis, because different interest rates, broadly speaking, tend to move together in practice. Macroeconomists do concern themselves with explaining why different assets yield different returns, however. Over the period 1926–1987, the real interest rate on U.S. Treasury bills averaged 0.5 percent. Other assets yielded much higher returns. Over the same period, the real return on long-term government bonds was 1.7 percent; the real return on long-term corporate bonds was 2.3 percent; the real return on common stocks was 8.8 percent; and the real return on small-company stocks was 14.2 percent. Such substantial differences in returns are naturally of interest to macroeconomists and financial economists. The returns on assets differ, in essence, for two reasons: first, because different assets have different risk characteristics and, second, because different assets have different terms to maturity. Explaining how the risk characteristics of assets affect their return is a topic that has given rise to a great deal of work in financial economics, and though much is understood, many puzzles also remain. The way in which the return on assets depends upon their term to maturity is known as the term structure of interest rates. As a preliminary to this, we consider the basic principles behind the pricing of a bond. A bond is an asset that is described in terms of three basic characteristics: its term to maturity; the coupon that it pays out each year; and its face value, which is the amount that it pays at maturity. Various assets exist that are special cases: zero-coupon bonds, as the name suggests, pay only at maturity; perpetuities (or consols) have an infinite term to maturity, or in other words, they simply pay a coupon every year, forever. The pricing of a bond in some ways resembles the pricing of a stock. Whereas the price of a stock is the present discounted value of the stream of dividends that it pays, the price of a bond is the present discounted value of the coupon payments and the face value. For simplicity, consider the case of a consol, in which case the price of the bond is given by PDV =⎜⎝⎛1+1r⎞⎟⎠ C +⎛⎝⎜1+1r⎟⎠⎞2C +⎛⎜⎝1 +1r⎞⎟⎠3C +  =⎧⎪⎨⎪⎩⎛ 1 ⎞ +⎛⎜1+1r⎞⎟⎠2 +⎜⎝⎛1+1r⎞⎟⎠3 +⎪⎭⎫⎪⎬C, ⎜⎝1+ r⎟⎠ ⎝ where C is the coupon payment. The sum of the infinite series in parentheses is simply 1/r, so we have C PDV = . r Equally, just as the holding return on a stock at time t depends upon the dividend that it pays next period and the change in the price (capital gain or loss) between this period and next period, so the holding return on a bond depends upon the coupon payment and the change in the price. The coupon payment, however, does not vary over time and so does not need to be predicted. Letting pt be the price of the bond and ht be the return on the bond, we have ht = C + pt+1− pt . pt pt If we set this return to be constant between now and the time of maturity (that is, set ht = r for every year), we can calculate the yield to maturity (r) of the bond. When people speak of the interest rate on a bond, they are referring to the yield to maturity. We have rpt = C + pt+1− pt  1  ⇒ pt =  1+r (C + pt+1). An equation like this holds for every period. By repeated substitution we can write the price of a bond at time t as6 confirming that the price of the consol is indeed simply the coupon divided by the interest rate. The important conclusion from this analysis is that bond prices and interest rates are inversely related: When bond prices fall, interest rates rise. The term structure considers how the interest rates on bonds of different maturities are related. The principal theory of the term structure of interest rates is known as the expectations theory. The basic idea of this theory is very intuitive: Agents should expect to gain the same return by investing in a long-term bond or by investing in a succession of short-term bonds. For simplicity, we restrict attention here to zerocoupon bonds. Thus, imagine an investor who can invest (at time t) in a two-year bond with annual interest rate rt(2), or who can instead invest in a one-year bond at the interest rate rt(1), and then take the proceeds and reinvest them in another one-year bond paying an interest rate rt+1(1) : The expectations theory of the term structure then states that 2 (1 +rt(2)) =(1 +rt(1))(1 +rt+(11)). As an approximation, we can write rt. In other words, the annual return on the two-year bond is approximately the average of this year’s and next year’s return on one-year bonds. This basic approach is easily extended to bonds of longer maturity. Investors making decisions this year do not know next year’s interest rate (rt+1(1)) with certainty, however. Thus, the term structure actually describes a relationship between the interest rate on a long bond and current and expected future interest rates on short bonds. The interest rate on long bonds relative to short bonds thus reflects agents’ expectations about future interest rates. To see this clearly, subtract rt(1) from both sides of the previous equation to get 6 We can again consider the more general case, in which this equation holds up to the period before the bond comes to maturity. The price when the bond comes to maturity is simply the face value: pt+M = F After repeated substitution we obtain so the price of a bond does indeed equal the present discounted value of the coupon payments and the face value. Note that zero-coupon bonds are particularly simple to price: ptZC = F M . (1+r) rt. According to this equation, the long rate will exceed the short rate if interest rates are expected to rise, and the short rate will exceed the long rate if interest rates are expected to fall. The expectations theory of the term structure explains theoretically why different interest rates tend to move together, since it shows that the interest rate on a long bond is simply an average of the interest rates on short bonds. The expectations theory does not explain another fact about the term structure, however: Long bonds generally have a higher return than short bonds. According to the expectations theory, the interest rate on long bonds is lower whenever investors anticipate a fall in interest rates, but it seems improbable that investors always expect interest rates to rise. The term structure as set out so far neglects that individuals may have a preference for shorter bonds because they dislike the risk associated with changes in the price (that is, capital gains or losses) of longterm bonds. There may therefore be a liquidity premium that must be paid to persuade individuals to hold long bonds. In this case, the long rate will exceed the short rate even if no change in interest rates is anticipated. Since this is what we observe in the data, it seems likely that liquidity premia are important. ADVANCED TOPIC 17-5 Asset Pricing IV: Bubbles, Excess Volatility, and Fads There are a number of dramatic historical incidents known as speculative bubbles, where the price of an asset rises and falls dramatically. One of the most famous is Tulipmania: In the Netherlands in the seventeenth century, certain rare varieties of tulip bulbs sold at extraordinarily high prices. “For example, a Semper Augustus bulb sold for 2,000 guilders in 1625, an amount of gold worth about $16,000 at $400 per ounce.” At the end of 1636 and the beginning of 1637, tulip bulb prices rose very rapidly and then collapsed suddenly; two years later, bulbs were selling for less than 0.1 guilder. Another famous example is the South Sea Bubble, during which the price of shares in the South Sea Company rose more than eightfold between January and July 1720 and then fell back to about their original level in the next three months. In these cases, it seems that the price of an asset changes not because of a change in the fundamentals but because investors demand the asset in the anticipation of future price rises, brought about in turn by more investors demanding the asset in the expectation of still further price rises, and so on. These incidents make many economists skeptical of the purported efficiency of financial markets. Further evidence of inefficiency is the observation that asset prices are much too variable to be explained in terms of changes in fundamentals. The economist Robert Shiller is a strong proponent of this view. Informal evidence of such excess volatility comes from the observation of major variation in asset prices, such as the stock market crash of October 1987 or the recent movements in real estate prices in some parts of the United States. Shiller and others have also provided more formal tests. Shiller’s argument rests on a property of rational expectations and some simple statistics. He noted that if the path of future dividends were known with certainty, then the perfect-foresight price of a stock would be given by the present value of the stream of dividends. The actual price that investors are willing to pay represents their prediction of this perfect-foresight price. If investors have rational expectations, then the actual price will be the best possible forecast of the perfect-foresight price. The perfect-foresight price then equals the actual price plus an unpredictable error term: ptpf = pt +ut. The term ut is the forecast error. If agents have rational expectations, this will be random and unpredictable. Shiller’s insight was that the actual price should be smoother than the perfect-foresight price, since some of the variation in the perfect-foresight price should be the result of forecast errors. But this is manifestly not true in the data: Actual stock prices are very volatile, and the perfect-foresight price is relatively smooth. Dividends turn out not to vary much, so it is hard to explain why stock prices exhibit large fluctuations. Researchers have also tried to see if the actual behavior of asset prices can be explained (after the fact) by the behavior of fundamentals. Richard Roll looked at the price of orange juice futures and argued that changes in these prices should be primarily caused by news of the Florida weather.7 In contrast to what market efficiency would suggest, he found that these prices varied much more than could be accounted for by weather news. Yet another way to look for inefficiency in asset markets is to see whether there are unexploited profit opportunities—ways to get something for nothing. Economists have looked for trading rules: simple rules that could be mechanically applied and that would earn “above average” profits. Karl Case and Robert Shiller suggested that a simple trading strategy could have exploited profit opportunities in the real estate market. Bruce Lehmann showed that a contrarian strategy of taking short positions in winners (selling stocks whose prices have recently increased) and long positions in losers (buying stocks whose prices have recently fallen) would have made a profit over the period 1962–1986. Such findings are intriguing but do not prove inefficiency. First, it has to be demonstrated that excess profits are large enough to overcome the transactions costs of implementing the rule. Second, identifying a profitable trading rule on the basis of past data is not a guarantee that it will work in the future. Third, as Donald McCloskey points out, such researchers “must answer the American Question: if you’re so smart, why ain’t you rich?” Of course, some providers of investment advice do get rich by advising the market. That some do succeed in outperforming the market is also apparently inconsistent with simple notions of market efficiency. There are many such advisors operating, however, and so we should expect that some should turn out to do well just by chance. The important question is, do we think that an investment service is likely to be able to beat the market in the future if we observe that it has succeeded in so doing in the past? On the basis of his research on different asset markets, Shiller argues that market efficiency cannot provide an adequate explanation of the behavior of asset prices. Shiller’s preferred theory is that fads and fashions play an important role in the determination of such prices. Progress in explaining the behavior of asset prices may require economists to move beyond naive models of rational behavior and pay much more attention to work in social psychology and sociology. Surveys seem to provide support for this view. For example, Shiller and co-researchers found that a majority of investors themselves explain the fall in the stock market in October 1987 in terms of “investor psychology” rather than movements in fundamentals. Surveys also reveal that speculative considerations were a major motivation for house purchases in areas experiencing real estate booms. Not all economists share Shiller’s view. Some believe that markets are basically efficient; apparent weaknesses in efficient-market models may just signal that we do not yet understand the fundamentals well enough. Others agree that markets are inefficient but suggest that cognitive psychological, rather than social psychological (that is, individually rather than socially based), explanations are more promising. ADVANCED TOPIC 17-6 Asset Pricing V: The Capital-Asset Pricing Model Different assets yield different returns. An important reason for this is that not all assets are equally risky. In general, investors care not only about the expected return on an asset but also about its risk characteristics. An important theory in financial economics, the capital-asset pricing model (CAPM), seeks to explain how the return on an asset is connected to its riskiness. The basic idea is that to induce agents to hold risky assets, it is necessary to compensate them with higher expected returns. But what exactly determines the riskiness of an asset? One natural supposition is simply that, the more variable is the return on an asset, the higher the return that it must pay. This is not quite right because it ignores the very important observation that individuals in general hold portfolios of assets and care only about their overall risk position, as determined by the variability of their portfolio. By holding a number of different assets (that is, by diversification), investors may end up with a portfolio that is less variable than any of the individual assets in their portfolio. The CAPM starts off by describing assets in terms of both their average return and their variance. For example, if asset A pays a return of 2 percent with probability 1/2 and 6 percent with probability 1/2, and asset B pays either 0 percent or 8 percent, each with probability 1/2, then their expected or average return is the same (4 percent), but asset B has higher variance. Now consider a portfolio that consists of 50 percent asset A and 50 percent asset B. Its expected return is simply the average of the expected return on the two assets, which in this case is (trivially) 4 percent. The variance of this portfolio, however, depends upon whether assets A and B tend to move in the same direction at the same time. As one example, suppose that the two assets always move together. That is, either both have a high return or a low return. Then the portfolio composed of the two assets will pay 7 percent with probability 1/2 and 1 percent with probability 1/2 (so, as asserted, it still pays an expected return of 4 percent). At the other extreme, suppose that the two assets always move in opposite directions, so either A pays 2 percent and B pays 8 percent or A pays 6 percent and B pays 0 percent. Then the portfolio pays 5 percent with probability 1/2 and 3 percent with probability 1/2. An agent holding this portfolio has diversified away some of the risk and so can still get a 4 percent expected return with lower variance. (In fact, it is possible to diversify away all the risk in this case. A portfolio consisting of two-thirds asset A and one-third asset B will guarantee a 4 percent return with no risk.) Similarly, we can imagine looking at all possible portfolios of all available assets and find those efficient portfolios that minimize the risk for any given return. If we graph this in terms of expected return and risk (variance), the set of efficient portfolios is then shown as the curved line in Figure 1. Now suppose that investors also have access to a risk-free asset, paying a certain return r. Investors can then obtain any combination of risk and return lying on the market line in Figure 1 by holding some combination of the risk-free asset and the market portfolio. If they want little risk, they will hold a lot of the risk-free asset and get a relatively low expected return (point X). If they are willing to accept higher risk for the sake of a higher return, they will hold more of the market portfolio (point Y). If they care a lot about return, they will borrow at the risk-free rate and invest in the market portfolio (point Z). Whereas investors could in principle hold any portfolio lying on or under the set of efficient portfolios, they would never want to do so. They could always get a higher return and/or lower risk by holding the market portfolio. The CAPM thus leads to the remarkable conclusion that all investors should hold the same portfolio of risky assets and take different risk positions simply by holding relatively more or less of the risk-free asset. Another important implication of the CAPM is that the key risk characteristic of an asset is not its individual riskiness but the extent to which it varies with the market. An asset that tends to move with the market portfolio possesses a great deal of undiversifiable risk; its price will tend to be relatively low. An asset that tends to move in the opposite direction to the market is valuable because it permits diversification and so will command a high price. (As an example, suppose in the earlier analysis that asset A is the market portfolio. Asset B would not be valuable if it tended to move with asset A but would be very valuable in the case where it tended to move in the opposite direction.) Assets with relatively low prices have relatively high expected returns, and vice versa. Thus, assets that move with the market are relatively risky (in the sense of undiversifiable risk), and so investors must be compensated with a relatively high expected return. Assets that move in the opposite direction to the market are valuable because of their risk characteristics and so have a lower expected return. In fact, the CAPM shows that the excess return (that is, the return above the risk-free rate) on an asset is related in a simple way to the tendency of an asset to move with the market. Specifically, we can write hi – r = βi (hm – r), where hi is the expected return on stock i, hm is the expected return on the market portfolio, r is the riskfree rate, and β measures the extent to which stock i covaries with the market. An asset that always moves with the market has beta = 1; an asset that is unrelated to the market has beta = 0; an asset that moves in the opposite direction to the market has a negative beta. The CAPM reveals that the excess return on an asset is simply proportional to its beta. The CAPM is an example of a model that had a significant effect on the world: Stock market analysts now routinely calculate the betas of different stocks. A generalization of the CAPM, the consumption-based CAPM, suggests that the price of an asset should depend not upon how it varies relative to the market, but on how it varies relative to consumption. People hold assets as a means of saving, and people save to smooth their consumption. People will therefore be particularly keen to acquire assets that offer a high return in times of relatively low consumption. A stock will be particularly valuable, for example, if it tends to have a high return in recessions, when consumption is relatively low. ADDITIONAL CASE STUDY 17-7 Financing Constraints in Japanese Firms A study by Takeo Hoshi, Anil Kashyap, and David Scharfstein provides some support for the idea that financing constraints are important. They studied two types of Japanese firms. One set, known as the keiretsu (industrial group), have close ties to large banks that finance their investment, so the firms find it relatively easy to borrow for new investment. A second group of independent firms have weaker ties to the banks and so have greater difficulty raising capital. These firms are likely to face significant financing constraints. Hoshi, Kashyap, and Scharfstein then investigate whether the amount of investment that firms carry out depends upon their liquidity, as measured by their cash flow and their holdings of short-term securities. Their main finding is quite striking: Investment by the independent firms seems to be much more dependent upon these variables than investment by the keiretsu firms. For example, according to their estimates, a given increase in cash flow would be associated with more than ten times as much extra investment at the independent firms as at the group firms. Financing constraints do seem to matter. ADDITIONAL CASE STUDY 17-8 Taxes, Babies, and Housing During the 1970s the United States experienced a nationwide boom in housing. The price of a new singlefamily home relative to the CPI rose 30 percent from 1970 to 1980. Economists do not know with certainty what caused the increase in housing prices during this period, but two hypotheses have been proposed. One hypothesis is that the rise in inflation and the failure of federal tax law to index for inflation caused an increase in housing demand. The federal income tax subsidizes homeownership in two ways: It does not require homeowners to pay tax on the imputed rent on their homes, and it allows homeowners to deduct mortgage interest when computing their taxable income. Because the nominal interest rate on mortgages rises when inflation rises, the value of this subsidy is higher at higher rates of inflation. Inflation and nominal interest rates rose substantially in the 1970s, which increased the tax benefits of homeownership. A second hypothesis is that the baby boom of the 1950s caused a rise in housing demand in the 1970s. Figure 1 shows the number of births each year from 1910 to 1989. Note that after World War II, births rose markedly—from 2.86 million in 1945 to a peak of 4.30 million in 1957. In the 1970s, the members of this large baby-boom generation began reaching adulthood and forming their own households. Therefore, the demand for housing grew rapidly, and housing prices rose. Source: U.S. National Center for Health Statistics. This baby-boom hypothesis suggested that the demand for new housing would fall during the 1990s. In the 1970s births fell substantially, reaching a low of 3.14 million in 1973. In the 1990s this small babybust generation reached adulthood. Some economists predicted that because of this slowdown in the growth of the adult population, real housing prices would fall during the 1990s. In fact, housing prices continued to rise in the late 1990s—following a pause early in the decade in the aftermath of the recession and the savings and loan crisis. Immigration and a rising stock market have been pointed to as reasons why the earlier prediction of declining house prices was wrong. The number of net immigrants during the 1990s was the highest of any decade during the twentieth century, while the number during the 1980s was the third highest. This influx of immigrants kept the home-buying population expanding faster than it otherwise would have. And rising stock-market wealth led homeowners to demand larger houses, putting further pressure on prices of homes. LECTURE SUPPLEMENT 17-9 The Tax Treatment of Housing Chapter 17 discusses the effects of tax laws on business fixed investment. Tax laws also have effects on residential investment. In this case, however, their effects are nearly the opposite. Rather than discouraging investment, as the corporate income tax does for businesses, the personal income tax subsidizes households to invest in housing. A homeowner can be viewed as a landlord who also rents her own house. But she is a landlord with special tax treatment. The United States does not tax her on the imputed rent (the rent she “pays” herself), yet it allows her to deduct mortgage interest. Thus, when computing her taxable income, she can subtract part of the cost of owning a home, but she does not have to add any of the benefit. The size of this subsidy depends on the rate of inflation because homeowners are allowed to deduct their nominal interest payments when computing taxable income. For example, when inflation and nominal interest rates rose sharply in the 1970s, the tax benefits of owning a home rose as well. When inflation and nominal interest rates fell in the 1980s and early 1990s, the tax benefits became smaller. Some economists have criticized the tax treatment of homeownership, arguing that, because of this subsidy, the United States invests too much in housing compared to other forms of capital. They support reducing the subsidy, perhaps by limiting the deductibility of mortgage interest and using the extra tax revenue to lower tax rates. The political response to this idea is mixed: Although voters prefer lower tax rates, homeowners are not ready to give up the mortgage interest subsidy that they have benefited from for so many years. CASE STUDY EXTENSION 17-10 The Importance of Inventories Inventory investment is a very small component of GDP. For example, in 2011, GDP was $17,421 billion, and gross private investment was $2,856 billion, whereas inventory investment was a mere $90 billion. Nevertheless, the behavior of inventories may be very important for macroeconomics, because fluctuations in inventories account for a substantial fraction of overall variation in GDP. (Remember that inventory investment can be negative.) Table 1 illustrates this dramatically. It shows the contribution of changes in inventory investment to the decline in real GDP during postwar recessions. In some cases, the change in inventory investment accounted for more than the overall decline in GDP. In other words, without the change in inventory investment, GDP would have risen, not fallen. The basic message is clear: We probably cannot hope to understand the business cycle unless we understand the behavior of inventories. Table 1 Inventory Investment and Postwar Recessions Cycle Peak to Trougha Change in Real GDPb Contribution from Inventory Investmentc Contribution as a Share of Decline in GDP 1948:4–1949:4 -1.43 -3.01 2.10 1953:2–1954:2 -2.36 -1.39 0.59 1957:3–1958:2 -2.85 -1.16 0.41 1960:2–1961:1 -0.28 -0.94 3.29 1969:4–1970:4 -0.11 -0.88 8.08 1973:4–1975:1 -3.04 -2.21 0.73 1980:1–1980:3 -2.12 -1.52 0.72 1981:3–1982:4 -2.47 -2.48 1.00 1990:3–1991:1 -1.32 -0.55 0.42 2001:1–2001:4d 0.47 -0.45 ------ 2007:4–2009:2 -4.12 -1.46 0.35 Source: Department of Commerce, Bureau of Economic Analysis and National Bureau of Economic Research (NBER). aPeaks and troughs correspond to NBER quarterly dates for business cycle recessions. bCumulative percent change from peak to trough. cCumulative percentage-point contribution to percent change in real GDP from peak to trough. dEven though inventory investment contributed negatively during this period, real GDP increased slightly in what was a relatively mild recession. ADDITIONAL CASE STUDY 17-11 Inventories and Production Smoothing The production-smoothing model does not do a good job of explaining inventories: Output and sales move closely together. The standard production-smoothing model suggests that production should be less variable (smoother) than sales. The economist Alan Blinder and others have documented that this is not the case. For example, Blinder compares the variance of sales to the variance of production for various manufacturing industries. For manufacturing industries as a whole, he finds that the ratio Variance Output Variance Sales equals 1.14. The ratio is greater than 1 for all but one (primary metals) of the 20 industries he considers and is as high as 2.4 in one industry (tobacco manufacturing). This evidence is not encouraging for the production-smoothing model.2 To see why, note first that production smoothing is based on the idea of a standard production function exhibiting diminishing marginal product. Think, for example, of a firm with a given stock of capital choosing how much labor to hire. Its production function takes the familiar form shown in Figure 1A. Diminishing marginal product implies that each extra unit of output requires more and more labor. The firm’s total cost curve is illustrated in Figure 1B: It shows total cost as a function of output. Now suppose that this firm faces seasonal demand. For simplicity, suppose it sells two units of output each month for six months of the year (say, the summer) and nothing for the remaining six months. The firm wishes to choose between two options: production smoothing, whereby it produces one unit every month and holds those goods in inventory when they are not in demand, and production bunching, whereby it produces goods as demanded (that is, it produces two units each month in the summer and no units each month in the winter). Figure 1B shows that the firm’s average cost under production smoothing is less than its average cost under production bunching. Providing that holding inventories is not too costly, the firm will wish to use inventories to smooth production. We arrive at a very different conclusion if the firm’s production function displays increasing returns to scale over some range. In this case, large-scale production may be more efficient than small-scale production. A production function exhibiting increasing returns (initially) is shown in Figure 2A; the associated total cost curve is shown in Figure 2B. As Figure 2B shows, it is now better for the firm to engage in production bunching rather than production smoothing. Greater variability of production than of sales can be reconciled with more standard assumptions on technology. One possibility is that firms face highly variable costs, although this is not very satisfactory because, as Blinder notes, it “comes perilously close to assuming the conclusion (production is variable because it is variable!).” Another possibility is that high demand in the present leads firms to anticipate even higher demand in the future. In this case, when firms see high demand, they want to build up inventories in anticipation of even higher sales in the future. The consequence is production bunching. This argument, however, does not explain why sales and production should be so closely related at a seasonal level also, since seasonal shocks are anticipated and transitory. Blinder and others have suggested that the S-s model of inventory adjustment may provide a better approach to understanding inventories, particularly in the retail and wholesale sectors. Think of a store that has to decide how to place orders for goods on its shelves. Placing orders is costly, and so the store wishes to place them only at intervals.44 It turns out that a good strategy for placing orders is to make sure that the stock of inventory lies between two numbers—an upper bound (S) and a lower bound (s). The store thus allows sales to run down its inventory until it has only s units left on the shelf and then places an order for S – s units to build its inventory back up to S. It then allows inventory to run down again until it reaches s, and so on. ADDITIONAL CASE STUDY 17-12 Production Smoothing and Coordination Failure A striking example of production smoothing occurred in the automobile industry in the 1930s. Up to 1935, production was extremely volatile in the auto industry: Most manufacturers introduced new models at the automobile shows in New York and Chicago in mid-January, while most demand was concentrated in the spring, with the onset of good weather. As a consequence, most production was concentrated in a few months in the early part of the year. This was costly both to auto manufacturers and to their suppliers. Because of the importance of the automobile sector (which accounted for about 5 percent of total industrial production), this also caused substantial employment volatility. The problem for individual manufacturers was that it was difficult for one producer to change the timing of new models. The automobile shows, where most manufacturers introduced their new models, provided essential publicity for the industry as a whole. Moreover, any manufacturer who released a model in advance of the auto show would reveal information about its new products to its competitors. The solution to this problem came in the form of the 1935 National Industrial Recovery Act, which was part of Roosevelt’s New Deal. This legislation made very specific provisions about practices in the auto industry. The act specified that new models should be introduced, and automobile shows held, in the fall. The industry did indeed make this switch in 1935, with the result that there were then two sales peaks, in the fall and the spring. Production was much smoother. The punchline to this story is that the National Industrial Recovery Act was actually ruled unconstitutional in May 1935. Yet the provisions that it contained were adopted by the auto manufacturers and persisted thereafter. It appears that the act helped the manufacturers coordinate on a new, better equilibrium that they could not achieve acting independently. ADVANCED TOPIC 17-13 The Multiplier-Accelerator Model The Nobel Prize–winning economist Paul Samuelson combined the idea of the accelerator relationship in investment and the multiplier from consumption to develop a model of cyclical behavior. Here is a simple example of his multiplier-accelerator model: Ct = C + c(Yt +Yt–1) It =β(Yt–1 –Yt–2) Yt = Ct + It +G. The first equation is the consumption function: Consumption depends upon current and last period’s income, perhaps for permanent-income reasons. The second equation is the accelerator relationship: Investment depends upon the change in GDP. The third equation is the GDP identity. If we substitute the first and second equations into the third, we find that Now suppose that there is a shock to this economy—for example, government spending is temporarily increased at time t = 1. Output rises in period one and rises further in period two as a result of the multiplier effect of increased consumption. The increase in output between periods one and zero encourages investment—the accelerator effect. This increase in investment, in turn, increases income and consumption. Output does not keep increasing at the same rate, however. As the change in output declines, investment falls. The economy may exhibit cycles as it adjusts to steady state. Figure 1 shows an example of this adjustment for the following model: C = 10; c = 0.45; β = 0.4; G = 10; ∆G = 5. LECTURE SUPPLEMENT 17-14 Additional Readings For a very readable discussion of theory and facts about inventory investment, see Alan Blinder and Louis Maccini, “Taking Stock: A Critical Assessment of Recent Research on Inventories,” Journal of Economic Perspectives 1 (Winter 1991): 73–96. Alan Auerbach discusses the effect of the 1986 tax reform on the cost of capital in A. Auerbach, “The Tax Reform Act of 1986 and the Cost of Capital,” Journal of Economic Perspectives 1, no. 3 (Summer 1987): 73–86. There is an extensive literature on asset pricing and stock market volatility. The Economist has a “Schools Brief” on the capital-asset pricing model (February 2, 1991): 72–73; discussions of this model can also be found in textbooks on finance. Richard Thaler’s “Anomalies” columns in the Summer 1987 (pp. 197–201) and Fall 1987 (pp. 169–177) issues of the Journal of Economic Perspectives contain discussions of various stock market puzzles. The Spring 1990 issue of the Journal of Economic Perspectives 4, no. 2, contains a symposium on “Bubbles.” The 1929 stock market crash is described by John Kenneth Galbraith in J.K. Galbraith, The Great Crash, 1929 (Harmondsworth: Penguin, 1961). There are a number of symposia on the 1987 stock market crash: “A Panel Discussion on the 1987 Stock Market Crash,” in S. Fischer, ed., NBER Macroeconomics Annual, 1988 (Cambridge, Mass.: MIT Press, 1988); Symposium on “Brady Commission Report on the Stock Market Crash,” Journal of Economic Perspectives 2, no. 3 (Summer 1988). Instructor Manual for Macroeconomics Gregory N. Mankiw 9781464182891, 9781319106058