CH-14-15 Aggregate Supply and the Short-Run Tradeoff Between Inflation and Unemployment – Macroeconomics : Chapter Notes

This Document Contains Chapters 14 to 15 CHAPTER 14 Aggregate Supply and the ShortRun Tradeoff Between Inflation and Unemployment Notes to the Instructor Chapter Summary This chapter summarizes current research on aggregate supply. It is divided into two parts. The first presents two prominent models of aggregate supply, emphasizing their common conclusion that output differs from the natural rate if prices differ from expected prices. The models are based on imperfections in goods markets that involve sticky prices or incomplete information. The second part moves from aggregate supply to the Phillips curve and uses this as a tool for discussing policy questions such as the costs of disinflation. Comments This chapter is one of the textbook’s real strengths. The synthesis of the two models of aggregate supply brings some unity to an area in macroeconomics in which there is a striking lack of consensus. From the point of view of explanation (though perhaps not of policy), it is very helpful to draw out the similarities between the Lucas imperfect-information model and new Keynesian theories of sticky prices. (Instructors who wish to pursue the policy implications of these models in more detail could use Supplement 14-7 on policy ineffectiveness.) Instructors can save time in this chapter by not presenting both models of aggregate supply and instead emphasizing the supply equation common to all the models. The material in this chapter requires about two lectures. Use of the Dismal Scientist Web Site Go to the Dismal Scientist Web site and download monthly data over the past ten years for the U.S. consumer price index, both for the overall index and the core index that excludes food and energy prices. Compute the 12-month inflation rate (i.e., December to December, January to January, etc.) to smooth out month-to-month volatility in inflation. Compare the measure of inflation for the overall index with the measure for the core index. Explain how one might interpret periods where these measures are significantly different from each other as periods during which supply shocks occurred. Discuss how these shocks may have shifted the short-run aggregate supply curve and Phillips curve for the economy. 313 Chapter Supplements This chapter includes the following supplements: 14-1 The Sticky-Wage Model 14-2 Real Wages over the Business Cycle 14-3 The Worker-Misperception Model 14-4 Anticipated and Unanticipated Money (Case Study) 14-5 Is Price Flexibility Stabilizing? 14-6 How Long Is the Long Run? Part Three 14-7 Policy Ineffectiveness 14-8 Did the NAIRU Decline in the 1990s? (Case Study) 14-9 Costs of Disinflation (Case Study) 14-10 The Unequal Costs of Disinflation (Case Study) 14-11 “The Poincaré Miracle” 14-12 Hysteresis and the Long-Run Phillips Curve 14-13 Unemployment in the United Kingdom in the 1980s 14-14 Additional Readings Lecture Notes Introduction The IS–LM model is a useful and versatile model of the economy in the short run when prices are fixed, and we could spend a long time analyzing its properties. Modern macroeconomics, though, treats IS–LM as only part of the explanation of short-run fluctuations: It is a theory of aggregate demand, which must be combined with aggregate supply to obtain a complete picture of the economy. The reason for this is that the assumption of complete price stickiness, with its implication of a horizontal short-run aggregate supply curve, is very unsatisfactory. Modern macroeconomics thus spends less time worrying about the niceties of the IS–LM model and instead focuses its attention on trying to explain aggregate supply in the short run. Much recent research in macroeconomics is on aggregate supply. Since we are looking at recent research, there is less consensus among macroeconomists as to what constitutes the best model of aggregate supply. Rather, there are a number of different competing models that nonetheless share many common elements and come to more or less the same conclusion. 14-1 The Basic Theory of Aggregate Supply This chapter introduces an upward-sloping short-run aggregate supply (SRAS) curve. As a result, changes in aggregate demand will now affect not only output in the short run, but also the price level. The upward-sloping SRAS curve arises due to “frictions” in the economy that we have thus far ignored. Our basic short-run aggregate supply equation is Y =Y +α(P− EP), where α is some number greater than zero and EP represents some expectation of the price level. This equation says that output may differ from its long-run natural rate if the actual price level turns out to be different from the price level that people anticipate. Note three implications of this equation. First, the aggregate supply curve slopes upward. Second, the position of the aggregate supply curve depends upon EP: The SRAS curve intersects the LRAS curve at a price level equal to EP, so a higher value of EP shifts the aggregate supply curve upward. Third, the parameter α measures how much differences between actual and expected prices influence output. If α = 0, the aggregate supply curve is vertical; if α becomes very large, the SRAS curve is almost horizontal, as in our earlier analysis. We consider two prominent models that explain why we might get such an SRAS curve: (1) the sticky-price model and (2) the imperfect-information model. These models highlight different “frictions,” and both contain some element of the truth. The Sticky-Price Model First, we turn to the sticky-price model. This model is the most widely accepted explanation for an upward-sloping SRAS curve. The model highlights how firms may sometimes set prices by long-term contracts with their customers or, in the absence of such contracts, hold prices constant simply because they don’t want to offend their customers with frequent price changes. Sticky prices also may result because it is costly to reprint catalogs and price lists or because nominal wages are sticky (perhaps due to social norms) and firms base their prices on the costs of production. The sticky-price model tells a very different story of how firms behave. Up to now, we have always thought of firms choosing how much output to produce, taking as given the price at which they can sell their output. If we really want to explain why prices may be sticky, we have to think about who actually sets prices and on what basis. In reality, firms generally set the prices at which they want to sell their output. To analyze this situation properly, we need models of imperfect competition in which firms have some monopoly power. We do not emphasize this point yet, but we return to it later. We suppose that a firm’s desired price is denoted by p. This price depends upon two things. First, the general price level (P): Think of the firm as having a desired real price in mind and then setting a nominal price based on the overall price level; or, essentially equivalently, think of the firm’s desired nominal price as depending on both the prices charged by its competitors and the cost of its inputs. A rise in the general price level causes these to rise and induces the firm to want a higher price. Second, the firm’s desired price depends upon the level of demand in the economy. A natural way to model this is to suppose that the firm would like to set a price equal to the general price level if demand is at its long-run level. If demand is higher, the firm wants a higher price. We can summarize this as p = P+a(Y –Y). Now suppose that there are two types of firms in the economy. One fraction (1 – s) has flexible prices and adjusts its prices to current economic conditions based on this equation. The other fraction, s, is characterized by sticky prices. That is, due to the costs of price setting or some other reason, they set a price in advance and keep it in place for some period of time. They must set a price on the basis of expectations about future economic conditions. This involves forecasting both the level of demand and the price level. We will suppose that their best forecast about the level of demand is that it is at its long-run level. This is consistent with the idea that there are random, unforecastable shocks to output that average out to zero. Then firms with sticky prices just set p = EP. The actual price level in the economy is the average of these two sets of prices: P = sEP+(1−s)(P+a(Y −Y)). We can solve this for P to obtain P = EP+(a(1−s) / s)(Y −Y). Supplement 14-1, This equation tells us, first, that if output is at its natural rate, the actual and expected price levels “The Sticky-Wage correspond. This makes sense. Sticky-price firms simply set their price equal to the expected Model” price level, whereas if output is at the natural rate, flexible-price firms set their price equal to the Supplement 14-2, actual price level. These together imply that the actual price level equals the expected price “Real Wages level. over the Business Cycle” This equation also shows that if demand is high, so that Y exceeds Y, then the actual price level will exceed the expected price level. High demand causes flexible-price firms to raise their prices. If all firms were flexible-price firms, then this story could not be consistent, since all firms would be trying to set a price above the average. In other words, if all firms had flexible prices, it would never be possible for Y to exceed Y . The behavior of the sticky-price firms serves to anchor the price level. The overall response of the price level to output shocks depends upon the mix of the two types of firms in the economy. When we rearrange this equation, we get an aggregate supply equation with α = s/((1 – s)a). The aggregate supply equation is flatter (output has a greater response to price changes) when s is big and a is small. When there are many sticky-price firms in the economy, the price will be near its expected level even when output is a long way from the natural rate. This story of aggregate supply is in some ways very different from the next one. While the aggregate supply curve looks essentially the same, the argument is not that unexpectedly high prices lead to high output but instead that when output is high, prices also tend to be high. Remember, however, that both the price level and the output level are endogenous variables. Supplement 14-3, “The Worker Misperception Model”” Supplement 14-4, “Anticipated and Unanticipated Money” Figure 14-1  Supplement 14-5, “Is Price Flexibility Stabilizing?”  Figure 14-2  Supplement 14-6, “How Long Is the Long Run? Part Three”  Supplement 14-7, “Policy Ineffectiveness An Alternative Theory: The Imperfect-Information Model The second model of aggregate supply, known as the imperfect-information model, assumes that markets clear. In other words, prices and wages are free to adjust so that supply equals demand in goods and labor markets. The market imperfection in this model is a temporary misperception about prices that causes the short-run aggregate supply curve to differ from the long-run aggregate supply curve. Robert Lucas developed the model in the early 1970s as a formalization of the framework sketched by Milton Friedman in his presidential address to the American Economic Association in 1968. For this reason, the short-run aggregate supply curve of this chapter is sometimes referred to as the Lucas supply curve. Consider a situation where producers are unable to distinguish perfectly between a shock to the price of their own output, which would lead them to alter their production, and a shock to the general price level, the best response to which would be no change in production. Given some positive price shock, a producer’s best response is somewhere between the two, with the exact decision on how much extra to produce depending upon how probable he regards each type of shock to be. More precisely, the model assumes that producers have some expectation of the general price level (EP). They then observe the price at which they can sell their output. If that price is unexpectedly high, they might conclude that the prices of all goods have increased (perhaps due to an increase in the money supply). In this case, they revise their estimate of the general price level and do not change their production. Alternatively, they might think that the demand for their output has increased, leading them to produce more. Any difference between the actual and expected prices could be due to either or both of these possibilities. Producers face a signal extraction problem. In such circumstances, the best response to an unexpected price increase is to increase output by some amount. The conclusion is by now familiar: If the price level is above the expected price level, output is above the natural rate. The imperfect-information model described in this chapter relies on confusion between relative prices and the aggregate price level, as originally developed by Nobel Prize–winning economist Robert Lucas. More recent work on imperfect-information models of aggregate supply, however, has emphasized the inability of people to fully incorporate information about the economy into their decisions rather than confusion about price changes. In other words, the “friction” that results in an upward-sloping SRAS curve is not the limited availability of information but, instead, the limited ability of people to use information that is available. This limited ability to process information leads price setters to respond only gradually to changes in macroeconomic conditions. Case Study: International Differences in the Aggregate Supply Curve The imperfect-information model is based on the fact that producers may be unable to distinguish perfectly between shocks to their own price and shocks to the general price level. If, however, shocks to the general price level predominate, then producers will tend to think that any given change in their price was probably due to such a shock and will not change output significantly. The converse is true if shocks to individual prices (real rather than nominal shocks) predominate. The model thus predicts that in countries where aggregate demand fluctuates a great deal, the aggregate supply curve will be steep (α will be small). Lucas examined data for many different countries and found evidence supportive of the theory: Aggregate demand had a smaller effect on output in countries where prices were highly variable. Chapter 5 pointed out that there is a well-documented though imperfectly understood correlation between the level and variability of inflation. Countries with highly variable inflation are thus countries where, on average, inflation is higher. When inflation is higher the costs of fixing nominal prices in advance will be greater, so individual firms’ prices will tend to be more flexible. Another explanation of Lucas’s finding is thus that, when aggregate demand is more variable, few firms will want to set prices in advance, and so the proportion of sticky-price firms in the economy is smaller. As a consequence, the aggregate supply curve is steeper. Implications The two models of aggregate supply differ in terms of the market friction they emphasize. The sticky-price model emphasizes that prices may not move to equilibrate supply and demand in the short run. The imperfect-information model emphasizes the role of information problems as an explanation of short-run fluctuations. These models together provide reason to believe that output will deviate from the natural rate whenever prices exceed their expected level. Putting this aggregate supply curve together with aggregate demand, we can see that an unexpected increase in aggregate demand will lead to an increase in both output and prices in the short run. In the long run, price expectations will adjust and the aggregate supply curve will move upward, implying that output will return to its natural rate at a higher price level. 14-2 Inflation, Unemployment, and the Phillips Curve Deriving the Phillips Curve From the Aggregate Supply Curve In 1958, the economist A.W. Phillips noted that there was an inverse relationship between the inflation rate and the unemployment rate. This relationship became known as the Phillips curve and is specified as π = Eπ – β(u – un) + ν, where un is the natural rate of unemployment and ν is a supply shock. To see how this is related to the SRAS curve, we first write the aggregate supply curve in terms of inflation: P = EP+(1/α)(Y −Y ) P− P−1 = EP− P−1+(1/ 2)(Y −Y ) ⇒π= Eπ+(1/α)(Y −Y ) (To be precise, we should think about the short-run aggregate supply curve as being specified in log terms.) Next, we can use Okun’s law to argue that positive output shocks are associated with unemployment’s being below the natural rate: π = Eπ – β(u – un). Finally, the modern Phillips curve is usually adjusted by recognizing that there will sometimes be shocks to the price level—most important, oil price shocks. Thus we get π = Eπ – β(u – un) + ν. The Phillips curve provides a useful way to discuss aggregate supply because it is framed directly in terms of two closely watched macroeconomic variables, namely, inflation and unemployment. FYI: The History of the Modern Phillips Curve In the early years after Phillips first noted the relationship between inflation and unemployment, economists thought that it might indicate a simple, stable tradeoff between inflation and unemployment: Lower unemployment could be achieved only at the cost of higher inflation, and vice versa. For a time, macroeconomic policy options were often discussed in terms of this hard choice. Macroeconomists’ understanding of the Phillips curve was greatly increased in the 1960s, when Milton Friedman and Edmund Phelps emphasized that the nature of the Phillips curve relationship should change when expectations of inflation change. The experience of the 1970s, when oil price shocks had a big impact on the economy, led economists to pay more attention to supply shocks when analyzing the Phillips curve. Supplement 14-8, “Did the NAIRU Figure 14-3 Decline in the 1990s?” Figure 14-4 Figure 14-5 Adaptive Expectations and Inflation Inertia Modern theories of aggregate supply, as summarized in the Phillips curve, confront a major difficulty. To use the Phillips curve, we need to have some idea of how people form expectations of inflation. This is a difficult problem, because expectations are not something we can directly observe. One possibility is that people look to the recent past; this is known as adaptive expectations: Eπ = f(πt–1, πt–2, . . .). A simple example is Eπ = πt–1. The consequence of this is inertial inflation; we observe rising prices in part because people expect rising prices. Two Causes of Rising and Falling Inflation The Phillips curve also tells us that inflation will rise if unemployment is below the natural rate or if there are price or cost shocks. Inflation that arises because of low unemployment, that is, (u – un) πt*) or output is above its natural level (Yt > Yt ), the real interest rate rises. And when inflation is below its target or output is below it natural level, the real interest rate falls. Unlike earlier chapters that focused on changes in the money supply as the policy instrument for the central bank, here the policy instrument is the interest rate. The implicit assumption here is that the central bank adjusts the money supply as necessary to achieve its target for the interest rate. Choosing the interest rate as the policy instrument is more realistic as it closely matches the practice of central banks around the world. Case Study: The Taylor Rule The Fed chooses a target for federal funds rate using two general guidelines: When inflation rises, the federal funds rate should rise, and when economic activity slows down, the federal funds rate should fall. Economist John Taylor has proposed a simple rule for the federal funds rate following these guidelines: Nominal Federal Funds Rate = Inflation + 2.0 + 0.5(Inflation – 2.0) + 0.5(GDP gap) where the GDP gap is the percentage by which real GDP exceeds its natural level. The federal funds rate responds to both inflation and the GDP gap when using this rule. For each percentage point by which inflation rises above 2 percent, the real federal funds rate rises by 0.5 percent; and for each percentage point by which real GDP rises above its natural rate, the real federal funds rate rises by the same amount—0.5 percent. If instead inflation falls below 2 percent or GDP falls below its natural rate, the federal funds rate falls accordingly. John Taylor’s monetary rule may be the rule that the Fed implicitly follows in setting policy. According to the Taylor rule, if inflation and output are low enough, then the nominal interest rate should be negative. This situation occurred during the financial crisis of 2008–2009. But the central bank cannot set a negative interest rate because people would simply hold money (which pays zero) rather than lend at a negative interest rate. The best that the central bank can do is keep the interest rate near zero, as the Fed did during the recovery from the crisis. By 2011, the Taylor rule began recommending an increase in the federal funds rate. But the Fed kept the rate close to zero. Some economists believe this was needed to make up for the period of time when a negative interest rate was suggested by the Taylor rule. Other economist, however, believe that the Fed has waited too long to raise interest rates as economic expansion gained speed. These economists are concerned that low interest rates might fuel future inflation. Figure 1: The Unemployment Rate and the Natural Rate of Unemployment in the United States. 15-2 Solving the Model The five equations presented above determine the paths of the model’s five endogenous variables: output Yt, the real interest rate rt, inflation πt, expected inflation Et–1πt, and the nominal interest rate it. Before using the model to analyze the economy’s response to economic shocks, we first describe the model’s long-run equilibrium. Table 15-1 The Variables and Parameters in the Dynamic AD–AS Model Endogenous Variables πt Inflation rt Real interest rate it Nominal interest rate Etπt+1 Real interest rate Exogenous Variables Yt Natural level of output πt* Central bank’s target for inflation εt Shock to the demand for goods and services υt Shock to the Phillips curve (supply shock) Predetermined Variable πt-1 Parameters Previous period’s inflation α The responsiveness of the demand for goods and services to the real interest rate ρ The natural rate of interest ϕ The responsiveness of inflation to output in the Phillips curve θπ The responsiveness of the nominal interest rate to inflation in the monetary-policy rule θY The responsiveness of the nominal interest rate to output in the monetary-policy rule The Long-Run Equilibrium Long-run equilibrium for the model is the situation in which there are no shocks and inflation is constant over time. Applying this to the five equations of the model gives output and the real interest rate equal to their natural values, inflation and expected inflation equal to the target rate of inflation, and the nominal interest rate equal to the natural rate of interest plus target inflation (Yt = Yt , rt = ρ, πt = πt*, Etπt+1 = πt*, it = ρ + πt*). The long-run equilibrium reflects the classical dichotomy whereby real variables are determined separately from nominal ones, and it exhibits monetary neutrality so that monetary policy does not influence real variables. The Dynamic Aggregate Supply Curve To analyze the economy in the short run, we need to derive two equations that are the analogues of the AD and AS equations of Chapter 14. The dynamic aggregate supply equation is the Phillips curve, with lagged inflation substituted for expected inflation: πt =πt−1+ϕ(Yt –Yt )+υt. (DAS) This equation gives rise to an upward-sloping schedule called the dynamic aggregate supply curve when plotted with inflation on the y-axis and output plotted on the x-axis. Its slope reflects the Phillips curve whereby high levels of economic activity give rise to high inflation. The curve is drawn for given values of lagged inflation, the natural level of output, and the supply shock. If these variables change, DAS will shift. Figure 2 The Dynamic Aggregate Demand Curve To derive the dynamic aggregate demand curve, start with the demand for goods and services equation and substitute for the real interest rate using the Fisher equation. Next, eliminate the nominal interest rate by using the monetary policy equation and substitute for expected inflation using the equation for inflation expectations. Finally, cancel terms and rearrange the equation to yield: Yt =Yt –[αθπ / (1+αθY )](πt –πt*)+[1/ (1+αθY )]εt. (DAD) This equation is represented as a downward-sloping schedule called the dynamic aggregate demand curve when plotted with inflation on the y-axis and output plotted on the x-axis. Its slope reflects the response of the central bank to inflation, whereby an increase in inflation leads to an increase in the nominal interest rate by more than the rise in inflation. This means that the real interest rate also increases, thereby reducing the quantity of goods and services demanded. The curve is drawn for given values of the natural level of output, the inflation target, and the demand shock. If these variables change, DAD will shift. Figure 3 The Short-Run Equilibrium The intersection of the dynamic aggregate demand curve and the dynamic aggregate supply curve determines the economy’s short-run equilibrium. These two relationships determine two endogenous variables (inflation and output in period t), given the five other exogenous (or predetermined) variables. These are the natural level of output, the target inflation rate, the demand shock, the supply shock, and the previous period’s inflation rate. The short-run equilibrium level of output can be less than, equal to, or greater than its natural level. In the long run, it will equal its natural level. Note that the short-run equilibrium rate of inflation becomes next period’s lagged inflation rate and so will influence the position of the dynamic aggregate supply curve in period t + 1. This link between periods is responsible for the dynamic patterns of adjustment of the economy in response to shocks or changes in policy. In other words, expectations of inflation in period t + 1, which determine the position of the dynamic aggregate supply curve in period t + 1, depend on the outcome of inflation in period t, providing a link between time periods. This continues into the future, with inflation in period t + 1 in turn determining expected inflation in period t + 2, etc. Figure 4 15-3 Using the Model We can use the model to assess the effects of change in the exogenous variables. To simplify the analysis, we assume the economy is initially at its long-run equilibrium. Long-Run Growth As discussed in Chapters 8 and 9, increases over time in the natural level of output, Yt , may occur due to population growth, capital accumulation, and technological progress. Both the DAD and DAS curves shift to the right by an amount equal to the increase in the natural level of output. Output increases by the same amount and inflation is unchanged. A Shock to Aggregate Supply Suppose that the aggregate supply shock variable ut increases to 1 percent for one period of time and then returns to zero. The DAS curve will shift up in period t by exactly the amount of the shock. The DAD curve will remain unchanged. Inflation rises and output falls in period t. These effects reflect in part the response of the central bank through its policy rule that leads to higher nominal and real interest rates, which in turn reduces demand for goods and services and pushes output below its natural level. Lower output dampens inflationary pressure, so inflation does not rise by the full extent of the supply shock. In periods following the shock, expected inflation is higher (since it depends on the previous period’s inflation), so the DAS curve does not return immediately to its initial position. Instead, adjustment occurs gradually with inflation declining and output rising, as the DAS curve gradually shifts to the right. Simulation analysis that uses realistic parameter values (see the FYI box) helps to illustrate the adjustment path for the economy over time in response to a supply shock, including paths for the nominal and real interest rates. The simulation analysis shows the phenomenon known as stagflation. Figure 6 Figure 7 FYI: The Numerical Calibration and Simulation The textbook uses simulation analysis to explore the adjustment of the economy to various shocks and changes in policies. Each period is best thought of as one year in length. The model is calibrated using numerical values for the parameters of the model and some of the exogenous variables. These are taken to approximate the actual U.S. economy and the policy rule John Taylor proposed, which broadly captures the behavior of the Fed. Graphs of the time paths of the endogenous variables after a shock are known as impulse response functions. A Shock to Aggregate Demand Suppose that the aggregate demand shock variable equals one for five periods and then returns to its normal value of zero. This positive shock might reflect a war that increases government purchases or a stock-market boom that raises wealth and consumption. More generally, a demand shock could represent any event that changes the demand for goods and services at given values of the natural level of output and the real interest rate. In period t, the DAD curve will shift to the right. The DAS curve remains unchanged in period t. Output and inflation both increase. As in the analysis of the supply shock, the effects work partly through the central bank’s policy rule. In response to higher output and inflation, the central bank raises nominal and real interest rates, thereby partly dampening the expansionary effect of the demand shock. Figure 8 In subsequent periods, expected inflation is higher, and so the DAS curve shifts upward continually, reducing output and increasing inflation. When the demand shock disappears in period t + 5, the DAS curve returns to its original position. But the DAS curve remains higher than its original level since expected inflation remains above its original value. As a result, the decline in demand pushes output below its natural level. The economy then gradually adjusts to its original position as inflation is slowly reduced and output expands. Figure 9 A Shift in Monetary Policy Suppose that the central bank lowers its target for inflation from 2 percent to 1 percent and keeps it at the lower value from then on. This will cause the DAD curve to shift to the left (and, to be exact, downward by one percentage point). Since the target for inflation does not enter the dynamic aggregate supply equation, the DAS curve does not shift initially. Output and inflation initially fall. Once again, the response of monetary policy is behind this outcome: A lower inflation target implies that actual inflation is now above target, so the central bank raises real and nominal interest rates. The higher real interest rate reduces the demand for goods and services, thereby lowering output below its natural level and lowering inflation along the initial DAS curve. Figure 10 Figure 11 With inflation lower, expected inflation falls as well, causing the DAS curve to shift downward. Over time, the process continues, with output rising and inflation falling until the economy reaches long-run equilibrium, with output at its natural level and inflation at the new target rate of 1 percent. The economy's adjustment to this change in monetary policy assumes that expectations are formed adaptively. But if the central bank’s announcement of the change in its target inflation Supplement 15-7, rate is credible, then people may lower their expectation about inflation immediately. In this “Inflation Inertia” case, where expectations are formed rationally based on all available information, the DAS curve will shift downward by the same amount and at the same time that the policy change shifts the DAD curve, and the economy will instantly reach its new long-run equilibrium. 15-4 Two Applications: Lessons for Monetary Policy The model developed in the previous sections can be used to motivate a discussion about the design of monetary policy. In particular, we can consider how the values of the parameters of the monetary policy rule influence the effectiveness of monetary policy. Figure 12 The Tradeoff Between Output Variability and Inflation Variability Consider the effect of a supply shock on the economy. As shown earlier, the initial response is for output to fall below its natural level and inflation to rise above the central bank’s target. But the extent to which output declines or inflation rises depends on the slope of the DAD curve. When the slope is steep, inflation rises relatively more and output declines relatively less, whereas when the slope is flat, inflation rises relatively less and output declines relatively more. Because the slope of the DAD curve depends on the parameters of the monetary policy rule, the central bank can affect the slope by choosing whether to respond more or less strongly to deviations from target inflation or the natural level of output. In particular, when 8r is large compared to 8Y, the DAD curve is relatively flat, and the central bank responds strongly to deviations from its inflation target by raising interest rates a lot to keep inflation contained; and it responds only weakly to deviations of output from the natural level. In this case, a supply shock has a relatively small effect on inflation and a relatively large effect on output. Alternatively, when the central bank responds weakly to deviations from its inflation target and strongly to deviations of output from its natural level, the DAD curve is relatively steep, and a Supplement 15-8, supply shock has a relatively large effect on inflation and a relatively small effect on output. The “Volatility and central bank thus faces a tradeoff between the variability of output and the variability of Growth” inflation. As discussed earlier, the central bank does not face a tradeoff between the level of output and the rate of inflation in the long run—since the classical dichotomy holds. But it does face a tradeoff between the variability of output and the variability of inflation. Case Study: Different Mandates, Different Realities: The Fed Versus the ECB The legislation that created the Federal Reserve gave it the dual mandate of stabilizing both employment and prices, whereas the European Central Bank (ECB) is charged with the primary objective of maintaining price stability, defined as inflation close to 2 percent over the medium term. These differences in mandates can be interpreted in our model as being reflected in different parameters in the monetary policy rule. For the ECB compared to the Fed, more weight is given to inflation stability and less to output stability. The events of 2008, when oil prices rose sharply and the world economy headed into recession, support this interpretation: The Fed lowered interest rates from 5 percent to a range of 0 to 0.25 percent, while the ECB cut interest rates by much less. In 2011, as the global economy recovered, the ECB began to raise interest rates while the Fed kept them close to zero. Apparently, the ECB was less concerned about recession and more concerned about keeping inflation contained, whereas the Fed was more concerned about limiting the recession. Over time, one might expect the ECB’s monetary policy to result in more variable output and more stable inflation in Europe compared to the United States. Assessing this prediction is difficult, however, because the ECB has only been in existence since 1998—not long enough to have sufficient data to determine the long-term results of its policy. Also, Europe and the United States differ in other ways besides the policy mandates of their central banks, and these may influence output and inflation independently from monetary policy. One example is that in 2010 several European nations, most prominently Greece, almost defaulted on their government debt. This crisis in the eurozone lowered confidence and aggregate demand around the world, but the effects were more severe in Europe than in the United States. The Taylor Principle Suppose that the central bank responded to a rise in inflation above its target by cutting the real interest rate. From the monetary policy equation, this would imply that the parameter 8r is less than zero: it =πt +ρ +θπ(πt –πt*)+θY (Yt –Yt ). Figure 13 In other words, the central bank would increase the nominal interest rate by less than the rise in inflation. This policy response will produce a positively sloped DAD curve. Now consider a oneperiod demand shock. This will shift the DAD curve to the right and initially lead to a rise in inflation and an increase in output. Higher inflation will lead to an increase in expected inflation for next period, which will shift the DAS curve upward. But with a positively sloped DAD curve, output rises further and remains above its natural level because the central bank lowers the real interest rate in response to higher inflation. Inflation becomes unstable, rising continually to higher levels. To prevent inflation getting out of control, the central bank must increase the nominal interest by more than the rise in inflation (8r must be greater than zero), thereby increasing the real interest rate, so that the DAD curve is negatively sloped. The requirement that the central bank respond to inflation by raising the nominal interest more than one-for-one is sometimes called the Taylor principle, after economist John Taylor, who highlighted it as a key consideration in the design of monetary policy. Case Study: What Caused the Great Inflation? Inflation during the 1970s in the United States reached high levels. Paul Volcker, who was appointed chairman of the Fed in 1978, instituted a change in monetary policy beginning in 1979 that eventually brought inflation under control. Volcker and his successor, Alan Greenspan, subsequently oversaw low and stable inflation for the next 25 years. Estimates of the monetary policy rule show that the parameter θπ was 0.72 during the Volcker-Greenspan regime after 1979 but –0.14 during the pre-Volcker era from 1960 to 1978. This suggests that monetary policy did not conform to the Taylor principle in the earlier period, possibly explaining why inflation got out of control. 15-5 Conclusion: Toward DSGE Models Advanced courses in macroeconomics develop a class of models known as dynamic, stochastic, general equilibrium (DSGE) models. The dynamic AD–AS model discussed in this chapter is a simpler version of these more advanced DSGE models. The model of this chapter illustrates how Supplement 15-9, the key macroeconomic variables (output, inflation, and real and nominal interest rates) respond “How Long Is the Long Run?” to shocks, interact with each other, and adjust over time. The model also provides insight into the design of monetary policy, highlighting the tradeoff that central banks face between variability in output and variability in inflation. And it suggests the important advice that central banks need to respond strongly to inflation so that it does not get out of hand. LECTURE SUPPLEMENT 15-1 How a Real Business Cycle Model Is Constructed The dynamic AD–AS model developed in this chapter can be used to analyze economic growth by considering how an increase in the natural level of output affects the economy. While this feature can be interpreted as incorporating a long-run growth path for the economy into a model of short-run fluctuations, it can also be interpreted as allowing for elements of the real business cycle approach (discussed in the appendix to Chapter 9) to play a role in short-run business cycle analysis. In this sense, the model of Chapter 15 can be viewed as a hybrid model that includes both Keynesian features that allow money to have short-run real effects and real business cycle elements that influence short-run fluctuations. The more advanced dynamic, stochastic, general equilibrium (DSGE) models mentioned in the text also exhibit these hybrid characteristics. This supplement and the several to follow discuss how real business cycle models are constructed and tested. Real business cycle models emphasize the role of technology shocks in driving short-run economic fluctuations. These models generally differ from other macroeconomic models, not only in their theoretical explanation of economic fluctuations, but also in the way they are tested with economic data. Typically, economists test a theory by ascertaining an implication of that theory for economic data and then applying statistical and econometric techniques to see whether the data are consistent with the theory. The approach of real business cycle theorists, however, has usually been to simulate the outcomes of their models and to compare those simulations with actual data. A simple illustration has been provided by the economist Charles Plosser.1 He considered the problem of a representative individual who has to make two decisions at any point in time.2 First, the individual must decide how to divide her time between leisure and working; and, second, she must decide how to divide her output between consumption and investment to increase future consumption. She makes these choices to maximize her expected utility (happiness), which depends upon her consumption and leisure now and at all times in the future.3 Plosser assumes that the individual also has access to a Cobb–Douglas production function: He thus assumes that the agent has one unit of time, so that 1 – Lt corresponds to leisure. The parameter β measures how much the agent values the present relative to the future, and the parameter η measures how much the agent values leisure relative to consumption. Y t = At KtαL1t−α. Here, At represents the possibility of random technology shocks. The first step in this model is calibration, or choosing values for the parameters of the model. In this case, Plosser has to choose values for capital’s share of output (α) and the depreciation rate of capital, as well as for parameters reflecting the individual’s preferences. These parameters are chosen on the basis of other information that we have about the economy.4 Next, Plosser solves the choice problems of the agent—in essence deriving a consumption function and a labor supply function. Finally, Plosser uses the Solow residual as a measure of technology shocks. Plosser then simulates the model. This means that he works out how this economy would behave under the assumption that the representative agent sees technology shocks as unpredictable and permanent. Plosser can then find the implied series for GDP, consumption, investment, employment, and real wages. Figures 1 to 5 show how Plosser’s simulations compare with the actual behavior of the U.S. economy. 1 C. Plosser, “Understanding Real Business Cycles,” Journal of Economic Perspectives 3, no. 3 (Summer 1989): 51–77. 2 If the economy is competitive (as Plosser assumes), then markets will allocate resources efficiently and we will not be badly misled by simply imagining that the economy consists of a single individual. 3 Specifically, Plosser assumes that the agent maximizes the following utility function: ∞ U =∑βt  log(Ct )+ηlog(1− Lt ) . t=0 He thus assumes that the agent has one unit of time, so that 1 – Lt corresponds to leisure. The parameter β measures how much the agent values the present relative to the future, and the parameter η measures how much the agent values leisure relative to consumption. 4 Specifically, Plosser chooses α = 0.42, β = 0.95, the depreciation rate = 0.1, and the parameter η such that Lt = 0.2 at all times. Source: Figures 1 to 5 from C. Plosser, “Understanding Real Business Cycles,” Journal of Economic Perspectives 3, no. 3 (Summer 1989): 51–77. Interpreting these figures is not easy, but they are quite striking. In particular, Plosser’s simulations for output and consumption seem to match up very well with the actual data, although the fit is worse for investment and labor market variables. These pictures indicate that a competitive economy hit by technology shocks can exhibit fluctuations that broadly resemble those experienced by the U.S. economy. A problem with this line of research is that there has been insufficient formal statistical analysis of what constitutes a good match between simulated results and actual data. Plosser’s simulations look as if they correspond to the U.S. data, but we cannot tell from inspection whether there is a good fit in a more formal statistical sense. Also, as discussed in Chapter 9 of the textbook, the Solow residual may not be a good measure of technological change. The methodology followed by Plosser is essentially that pursued by most real business cycle theorists, except that they do not usually assume a specific series (such as the Solow residual) for technical change. Instead, they simply suppose that there are random shocks to the technology drawn from some statistical distribution. Rather than running just one simulation, real business cycle theorists simulate their models many times over. By doing this often enough, they can discover the broad patterns that their models imply for the data (for example, the correlation between output and consumption). They then compare these patterns with those observed in actual data. 5 Much modern research in macroeconomics examines refinements of this basic model in an attempt to improve the match between the models and the data. Some researchers are pursuing versions of the model that include the sort of imperfections emphasized by new Keynesian economists. As a result, many economists are hopeful that a synthesis of real business theory and new Keynesian economics is starting to emerge through the development of DSGE models in which money has effects on real variables in the short run alongside the effects of technol See also Supplement 8-3, “Does the Solow Model Really Explain Japanese Growth?” for another use of a real business cycle model. That supplement discusses Christiano’s simulation of a neoclassical growth model to investigate the hypothesis that Japanese saving behavior is explained by post–World War II reconstruction. LECTURE SUPPLEMENT 15-2 The Microeconomics of Labor Supply Many economists are unconvinced that real business cycle theory can adequately explain fluctuations in employment. To pursue this further, we start from two features of this theory: first, prices are assumed to be flexible; and, second, shocks to technology are the driving force behind economic fluctuations. Since prices are flexible, it follows that the labor market is always in equilibrium, so labor demand always equals labor supply. Technology shocks affect the marginal product of labor and so cause the demand for labor to shift. Looking at the effects of shifts in labor demand in Figures 1A and 1B, we see that steep labor supply implies little variation in employment and large variation in the real wage; whereas if labor supply is flat, then real wages would be relatively stable and employment would vary substantially. The data exhibit much employment fluctuation and little real-wage variation. It follows that, to explain the data, real business cycle theories need either a relatively flat labor supply curve or an explanation of why technology shocks might also shift labor supply. We consider each in turn. Neither theory nor the data provide a great deal of support for a flat labor supply curve. An individual’s labor supply decision is based on the choice between leisure and goods. Individuals have a certain amount of time at their disposal, which they can either take as leisure or use for work to earn income with which to buy goods. The real wage, w, is the relative price of leisure in terms of goods. The higher the real wage, the more goods must be forgone for an extra hour of leisure. We illustrate this in the standard microeconomic manner in Figure 2. We suppose that the individual has 24 hours to allocate between working and leisure. At one extreme, she could not work at all and take all her time as leisure. At the other extreme, she could work all 24 hours, have no leisure time, and consume 24w worth of goods. The line connecting these two points is the budget line; any point on this line is a feasible combination of leisure and goods. The optimal combination of goods and leisure is found where the indifference curve is tangent to the budget line. Now suppose that the real wage rises to w. We can see from Figure 2 that although leisure has now become more expensive, the individual may increase (case A) or decrease (case B) her supply of labor. The reason is that changes in the real wage generate income and substitution effects that act in opposite directions. The substitution effect encourages people to work more (that is, consume less leisure) when the wage goes up. A rise in the real wage, however, increases the income from working, allowing the individual to consume more leisure. Thus, the effect of an increase in the real wage on labor supply is theoretically ambiguous. It is perhaps not surprising that the data show that changes in the real wage do not have strong effects on labor supply. In terms of our original diagrams, therefore, the labor supply curve is steep. Contrary to the data, we would expect to see large changes in the real wage and small changes in employment if the economy is competitive and characterized by changes in labor demand. We observe a larger change in employment and a smaller change in the real wage if technology shocks affect both labor demand and labor supply in the same direction. This can occur if the interest rate changes or if there is a temporary change in the real wage. For example, if the real wage is high in the present but expected to be low in the future, workers might prefer to work very hard when the wage is high and take much leisure time when the wage is lower. To put the same point a slightly different way, labor supply might be very responsive to short-run fluctuations in the real wage, even if it is not responsive to long-run changes. Similarly, if the interest rate is higher, working today looks relatively attractive. We can illustrate this in terms of the labor market by putting the current real wage on the axis and noting that changes in the expected future real wage or the interest rate shift the labor supply curve, as in Figures 3A and 3B. An increase in the future real wage (wt + 1) makes the current supply of labor less attractive and so causes the labor supply curve to shift inward. An increase in the interest rate is like a decrease in the future real wage and so shifts the labor supply curve outward. Now, consider a temporary shock to labor demand (caused perhaps by a temporary shock to the technology). This shock does not affect the future real wage and so leads to a relatively large change in employment. Such a shock is unlikely to have a large effect on the interest rate. On the other hand, a permanent (positive) shock to labor demand leads to expectations of higher real wages in the future, causing the labor supply curve to shift in. This results in a relatively small change in employment. A focus on temporary changes in the real wage thus allows real business cycle theory to explain fluctuations in employment while recognizing that labor supply need not be very sensitive to real wages in the long run. Microeconomic analyses of individual labor supply, however, are still not very supportive of strong intertemporal substitution effects of this kind—that is, they do not indicate that labor supply is very responsive to temporary real wage changes or to changes in the interest rate. LECTURE SUPPLEMENT 15-3 Quits and Layoffs Job separations can occur either because workers voluntarily quit their jobs or because they are laid off (or fired with cause). We can thus write s = q + l, where s is the separation rate (see Chapter 7), q is the quit rate, and l is the layoff rate. Theories of intertemporal substitution argue that employment is lower in recessions because the real wage (or the interest rate) falls and workers are unwilling to work at this lower wage. Such an explanation suggests that quits should be an important component of flows from employment to unemployment, and also that quits should be higher in recessions. The data reveal, however, that layoffs are much more important than quits in explaining flows into unemployment. Data suggest that less than 15 percent of the unemployed become unemployed as a result of quitting their job. For example, unemployment in 2005 was 7.6 million. Of these, 3.7 million (48 percent) were unemployed as a result of layoffs, and 0.7 million (9 percent) were new entrants into the labor force. Of the remainder, 2.4 million (31 percent) had been previously employed and were reentering the labor force after a spell of nonparticipation. Only 0.9 million (12 percent) were job leavers—that is, individuals who quit their jobs voluntarily. Finally, the data indicate that quits are procyclical and layoffs are countercyclical. These data do not support the belief that employment fluctuations over the business cycle are the result of voluntary shifting of labor. LECTURE SUPPLEMENT 15-4 Involuntary Unemployment and Overqualification Some economists distinguish between two types of unemployment: voluntary and involuntary. According to the usual definition, someone is voluntarily unemployed if, at the existing wage, she does not think it worthwhile to work. A person who is involuntarily unemployed would like to work at existing wages but cannot obtain a job. Other economists argue that the idea of involuntary unemployment makes no sense. After all, they suggest, an unemployed investment banker or neurosurgeon could always get a job flipping hamburgers or waiting tables. So how can we distinguish between involuntary and voluntary unemployment? Robert Lucas expands on this argument as follows: Nor is there any evident reason why one would want to draw this distinction. Certainly the more one thinks about the decision problem facing individual workers and firms the less sense this distinction makes. The worker who loses a good job in prosperous times does not volunteer to be in this situation; he has suffered a capital loss…. Nevertheless the unemployed worker at any time can always find some job at once…. Thus there is an involuntary element in all unemployment, in the sense that no one chooses bad luck over good; there is also a voluntary element in all unemployment, in the sense that however miserable one’s current work options, one can always choose to accept them. Truman Bewley, an economist at Yale University, interviewed a large number of businesspeople to learn more about their decisions about hiring workers. His findings suggest that it may not be quite so easy for unemployed workers to find jobs, even at lower wages : Cannot workers find jobs immediately simply by accepting sufficiently low pay? Perhaps the clearest regularity of the survey was that large classes of unemployed workers find it very difficult to obtain work paying substantially less than what they earned before, unless they take temporary jobs or low-paying jobs in the secondary labor market. Most employers offering good permanent jobs shun workers who earned significantly more previously, significantly meaning 20–30 percent more. Employers label such workers as overqualified and fear that they will be discontent, be a threat to their supervisors, and above all, will leave as soon as they find better jobs. Note that Bewley’s findings do not completely contradict Lucas’s argument. They suggest that the unemployed investment banker could indeed get a job flipping hamburgers. But they also suggest that this unfortunate investment banker probably cannot do much better. ADVANCED TOPIC 15-5 Why Technology Shocks Are So Important in Real Business Cycle Models In any competitive flexible-price model, such as those espoused by real business cycle theorists, labor market clearing implies that the real wage must equal the marginal product of labor. As explained in Chapter 3, the marginal product of labor gives the firm’s demand for labor and depends upon the amount of capital and labor that firms possess. In particular, we expect to see diminishing marginal product: the marginal product of labor will be lower when employment is higher, absent other complications. We write MPL(K, L) = W/P. We know that employment is procyclical—not surprisingly, employment is higher in booms and lower in recessions. Other things being equal, we would expect to see the marginal product of labor falling in booms, and hence, if the economy is competitive, we would expect to see the real wage also being lower in booms. But we also know from Case Studies 14-1 and 14-2 that the real wage is actually mildly procyclical—real wages are higher in booms and lower in recessions. It follows that if we are to reconcile a procyclical real wage with diminishing marginal product of labor in a competitive model, other things are not equal. Something must happen in booms to raise the marginal product of labor even though employment is higher. Since the capital stock changes only slowly and does not vary in any significant way over the business cycle, the only possibility is that the marginal product of labor is higher in booms because of technological improvements. This is why technology shocks are an essential ingredient of real business cycle models. If the economy is not competitive, these issues need not arise. First, the demand for labor may depend upon other factors. For example, when prices are sticky, firms may demand less labor in recessions because they cannot sell all their output (whereas in a competitive model with flexible prices, firms can always sell as much as they want at the market price). As another example, suppose that the economy is not perfectly competitive but instead exhibits imperfect competition. We can rewrite the earlier equation as P = W/MPL. This says that, in a competitive economy, the price of output equals the marginal cost of production (since the wage is the cost of a unit of labor and the marginal product of labor gives the amount of output contributed by the last unit of labor). Under imperfect competition, however, firms set prices as a markup (m) over marginal cost: P = m × (W/MPL) ⇒ MPL/m = (W/P). In this case, the real wage can be procyclical even if the marginal product of labor is countercyclical, provided that the markup is also countercyclical. That is, if markups are higher in recessions, then MPL/m will be lower, and so the real wage may be lower. There are reasons for believing that markups may indeed be countercyclical. One reason why markups may fall in booms is that higher profits when the economy is booming cause more firms to enter an industry. The more firms in the industry, the closer it is to being competitive, and so the lower is the markup. Another possibility is that imperfect competition reflects collusion among firms, and firms maintain such collusion by a threat to increase output if other firms cheat. In a boom, demand is high, so the potential gain from cheating is greater and firms can sustain less collusion. Mark Bils investigated the behavior of marginal cost and price over the course of the business cycle. He found that marginal costs are strongly procyclical but prices do not vary much over the business cycle. His evidence suggests that the markup—the difference between price and marginal cost—is countercyclical. Julio Rotemberg and Michael Woodford investigated a real business cycle model with some imperfect competition and countercyclical markups. They carried out simulations and argue that they were better able to match the U.S. data by their inclusion of imperfect competition. This may be an encouraging route for synthesis between real business cycle and new Keynesian theories. Once we introduce imperfect competition, however, there is no longer a presumption that fluctuations are efficient and there may be a case for government intervention to stabilize the economy. ADVANCED TOPIC 15-6 Real Business Cycles and Random Walks Real business cycle theory provides a challenge to the traditional explanation of macroeconomic fluctuations. One reason why this theory has been so influential is the work of two economists, Charles Nelson and Charles Plosser. In an important article published in 1982, Nelson and Plosser argued that there is evidence to suggest that U.S. GDP may follow a random walk. That is, they suggested that the behavior of real GDP over time could be described by the equation Yt = Yt–1 + ut , (1) where ut is a random shock that is zero on average. This equation states that the best prediction of GDP this year is whatever value it had last year. The conventional view of macroeconomic fluctuations is that the behavior of GDP over time can be decomposed into a long-run natural-rate or trend component and a short-run cyclical component. This approach underlies the models used in the textbook: The Solow growth model explains the long-run behavior of the economy and the aggregate demand–aggregate supply model explains short-run fluctuations. In this view, shocks to the economy will push it away from the natural rate only temporarily; the economy always has a tendency to revert to the natural rate. But the Nelson–Plosser finding challenges this characterization. If GDP does follow a random walk, then shocks to output have permanent effects. To see this, suppose that at some time (t = 0), GDP is at the value Y0, and that at t = 1 there is a oneunit shock to GDP (u1 = 1). Suppose also that there are no further shocks (u2 = u3 = . . . = 0). Then Y1 = Y0 + 1. Now Y2 = Y1 + u2 = Y1 = Y0 + 1. Similarly, Y3 = Y0 + 1, and so on. The shock to GDP in period 1 persists forever. Following this shock, our best prediction about GDP is that it will forever be one unit higher (Figure 1). The observed fluctuations in GDP, according to this theory, are then fluctuations in the natural rate of output, not cyclical fluctuations of output around the natural rate. Whereas the traditional theory suggests that technological progress is a relatively smooth and gradual process, real business cycle theory suggests that technological progress is irregular and a source of fluctuations. Indeed, if this real business cycle characterization of the data is accurate, then the traditional decomposition of output into cycle and trend does not really make sense. If GDP does not follow a random walk, then the conclusion is very different. Suppose, for example, that the behavior of GDP can be described by the equation Yt = 0.9Yt–1 + ut. (2) Then, if we carry out the same experiment, we find that the shock raises output by 1 at time t = 1, as before. Next period, however, output is 0.9 higher as a result of the shock. In period 3, output is only 0.81 (= 0.92) units higher, and so on. In other words, the impact of the shock on GDP gradually dies out. In this representation, shocks to the economy are temporary, not permanent, and output does tend to return to the natural rate following a shock (Figure 2). So, if the Nelson–Plosser result is right, and GDP can be well described by a random walk, we need to think in terms of models where shocks have permanent effects. In terms of standard aggregate demand– aggregate supply models, this suggests that real or supply shocks, such as to technology, govern the behavior of GDP; aggregate demand shocks do not have permanent effects on output in such models. Demand shocks may, however, have permanent effects on GDP in other models such as the hysteresis models discussed in Chapter 14. Steve Durlauf, however, points out that if GDP follows a random walk, it is also consistent with a world in which coordination failures are important. In this case, demand shocks might push the economy from one equilibrium to another. Unfortunately, it is very hard to distinguish in the data between equations (1) and (2), and so we simply are not sure whether the random-walk characterization is accurate. Very different theories will generate very similar predictions for the behavior of GDP. For example, a world with demand shocks and very sticky prices is one in which shocks would exhibit a great deal of persistence, so GDP might appear close to a random walk. On the basis of GDP data alone, it is nearly impossible to distinguish between this economy and an economy governed by real shocks. Modern macroeconomics is making progress toward a synthesis in which it is recognized that both demand and supply shocks have important effects on output. In this view, the natural rate of output grows irregularly, as suggested by real business cycle theory, rather than exhibiting the smooth change of the Solow growth model. Nevertheless, demand shocks may still cause the actual level of GDP to differ from the natural rate and so may be an additional source of variability in GDP. In principle, in such a world, there is still room for stabilization policy to eliminate inefficient cyclical fluctuations. Eliminating all fluctuations is no longer desirable, however, since some variation in GDP is an efficient response to technology shocks. Although many economists doubt that real business cycle theory completely explains economic fluctuations, most might agree that it teaches the important lesson that some variation in GDP is to be expected and is indeed desirable in a well-functioning economy. LECTURE SUPPLEMENT 15-7 Inflation Inertia The Phillips curve used in the dynamic AD–AS model of Chapter 15 can be derived under the assumption that all firms have the ability to set prices and some of those firms set their prices one period in advance. As shown in Chapter 14, this assumption implies a Phillips curve that relates period t inflation to the period t – 1 expectation of period t inflation and the gap between actual and the natural level of output. Introducing adaptive expectations then allows derivation of the DAS curve, which relates period t inflation to period t – 1 inflation and the deviation in output from its natural level. The effect of lagged inflation in the DAS curve is responsible in the model for the gradual adjustment of inflation in response to shocks. A more sophisticated approach, known as staggered price setting, assumes that firms all set prices in advance for two periods, with half of the firms setting prices in any given period. Staggered price setting makes the overall level of prices adjust gradually, even when individual prices adjust frequently. In other words, the price level will adjust fully to an increase (decrease) in aggregate demand only after a period of time during which output exceeds (falls short of) its natural rate. But a surprising implication of these new Keynesian models of staggered price setting under rational expectations is that inflation—the percent change in prices—does not exhibit inertia. Instead, inflation is expected to decline when output is above its natural rate and vice versa. The reason for this result is that when price-setters fix a price for the current and future periods, they consider not only today’s overall price level but also the price level expected to prevail in the future. The resulting Phillips curve expresses inflation as a function of next period’s inflation and the current output gap. Accordingly, a declining path for inflation is associated with output above its natural rate. Evidence for the United States and many other countries contradicts this implication and supports the view that inflation is highly persistent. Periods of disinflation across countries are overwhelmingly periods when output is below normal. And estimates of the inflation process for the United States find that lagged inflation helps explain current inflation. Various ways of reconciling new Keynesian models of price dynamics with evidence of inflation inertia have been proposed. These include adding delays in price adjustment, incorporating some backward-looking price-setters, indexing fixed prices to overall inflation between adjustments, and introducing more complex dynamics in costs or markups. Greg Mankiw and Ricardo Reis have suggested changing the basic framework from one with “sticky prices” to one with “sticky information.” Instead of assuming full information with staggered price setting, Mankiw and Reis assume firms can always adjust prices but are limited by the cost of obtaining and processing information. As a result, firms may choose a path for their prices that is set until the next time they update their information. The result leads to a Phillips curve in which past inflation affects current inflation and in which disinflations are associated with below-normal output. One drawback of the Mankiw–Reis approach is that it does not allow a role for fixed prices, despite evidence of their importance in the economy. In addition, the sort of predetermined paths that firms choose in their model do not appear to be widespread in the economy. Furthermore, fixed prices appear essential for explaining why shifts in aggregate demand have smaller and shorter-lasting effects in high-inflation economies and why the announcement in advance of disinflation policies doesn’t measurably affect the output costs of disinflation. Most likely, a complete framework for explaining inflation dynamics will require both fixed prices and predetermined price paths. ADDITIONAL CASE STUDY 15-8 Volatility and Growth Garey Ramey and Valerie Ramey investigated the connection between the growth and the volatility of GDP in a number of different countries. They wished to find out if long-run growth and short-run volatility were related. As a matter of theory, growth and volatility could be directly or inversely related. For example, large fluctuations in output might make firms reluctant to commit to irreversible investment, implying that growth would be lower in countries with highly variable output. Conversely, consumers in a relatively uncertain world might save a lot, which could lead to higher growth. Figure 1 shows the relationship between volatility and growth in OECD countries, measured over the period 1952–1988. There is a strong negative relationship: Countries with highly variable output tend to be countries that grow more slowly, and conversely. One implication of Ramey and Ramey’s findings is that the benefit of reducing business cycle fluctuations might therefore be larger than is commonly supposed: Stabilization of the economy in the short run might help promote growth in the long run. Source: G. Ramey and V. Ramey, “Cross-Country Evidence on the Link Between Volatility and Growth,” American Economic Review 85, no. 5 (December 1995): 1143. LECTURE SUPPLEMENT 15-9 How Long Is the Long Run? Part Four Macroeconomists traditionally decompose the overall behavior of GDP through time into its long-run growth (or trend) and its short-run fluctuations (or cycle). That is the approach followed in the textbook. Chapters 3, 6, 8, and 9 explain the determination of the natural level of output at a point in time and show how the natural level of output grows through time as the economy’s resources and technology change. Chapters 10 to 14 explain how actual GDP may differ from the natural level in the short run because of shocks to aggregate demand combined with an upward-sloping aggregate supply curve (as a result of price stickiness or information imperfections). Thus, Chapters 3, 6, 8, and 9 explain the trend growth of GDP, whereas Chapters 10 to 14 explain the business cycle. The simple dynamic model presented in Chapter 15 incorporates elements of both short-run business cycle fluctuations and long-run economic growth into a unified framework. It does so by allowing for growth over time in the natural level of output within a model that has sticky prices in the short run. Economists have developed much more sophisticated models, known as dynamic, stochastic, general equilibrium (DSGE) models, in which this traditional decomposition between trend and cycle can be misleading. Both the “short-run” fluctuations in output and the “long-run” growth of output are, according to this view, in part manifestations of the same phenomenon—the response of the economy to technology shocks. To put it another way, output sometimes fluctuates because the natural level of output fluctuates. But it may also fluctuate because of shifts in aggregate demand arising from changes in the money supply when prices are sticky. Hence, DSGE models are hybrids that combine both Keynesian elements and real business cycle elements into a single approach. LECTURE SUPPLEMENT 15-10 Additional Readings The Summer 1989 issue of the Journal of Economic Perspectives 3, no. 3, contains two articles on real business cycle theory: one by Charles Plosser, a proponent of the theory, “Understanding Real Business Cycles,” pages 51–77; and one by Greg Mankiw, who is more skeptical, “Real Business Cycles: A Keynesian Perspective,” pages 79–90. A useful, but more technical, survey is B. McCallum, “Real Business Cycle Models,” in R. Barro (ed.), Modern Business Cycle Theory (Cambridge, Mass.: Harvard University Press, 1989). The Fall 1986 issue of the Federal Reserve Bank of Minneapolis Quarterly Review 10, no. 4, contains a debate on the topic between Edward Prescott and Lawrence Summers. Rodolfo Manuelli’s introduction is also very useful. Much work on real business cycles has focused on the labor market. For a survey, see G. Hansen and R. Wright, “The Labor Market in Real Business Cycle Theory,” Federal Reserve Bank of Minneapolis Quarterly Review 16, no. 2 (Spring 1992). There are a number of good surveys of the current state of macroeconomics, including Robert Gordon, “What Is New-Keynesian Economics?” Journal of Economic Literature 28 (September 1990); Bennett McCallum, “Post-War Developments in Business Cycle Theory: A Moderately Classical Perspective,” Journal of Money, Credit, and Banking 20 (August 1988); Greg Mankiw, “A Quick Refresher Course in Macroeconomics,” Journal of Economic Literature 28 (December 1990): 1645–60; Greg Mankiw and D. Romer, “Introduction,” in G. Mankiw and D. Romer, eds., New Keynesian Economics (Cambridge, Mass.: MIT Press, 1991). The Mankiw and Romer volumes also contain many of the important papers on new Keynesian economics. The Journal of Economic Perspectives 7, no. 1 (Winter 1993), contains a symposium on “Keynesian Economics Today” that includes articles by avowed new Keynesians David Romer, Bruce Greenwald, and Nobel Prize-winner Joseph Stiglitz; self-described old Keynesian and Nobel Prize-winner James Tobin; and Robert King, who is skeptical of the new Keynesian approach. Instructor Manual for Macroeconomics Gregory N. Mankiw 9781464182891, 9781319106058