In Chapter 7 we obtained the following aggregate demand relation, when the central bank kept the money stock constant (equation [7.3]), between output and the real money stock, government spending and taxes based on equilibrium in goods and linancial markets:
У, = (9-7)
where we have added time indexes—which we didn't need in Chapter 7 but will need in this chapter.
To focus on the relation between the real money stock and output, we will ignore changes in factors other than real money here, and write the aggregate demand relation simply as
Y, = y^T (9-8)
' i
where у (the Creek lowercase gamma) is a positive parameter. This equation states that the demand for goods, and thus output, is simply proportional to the real money stock. This simplification will make our life easier. You should keep in mind however, that behind this relation hides the mechanisms you saw in the IS-I.M model:
• An increase in the real money stock leads to a decrease in the interest rate.
• The decrease in the interest rate leads to an increase in the demand for goods, and so to an increase in output.
Equation (9.6) gives a relation between levels—the output level, the level of money and the price level. We need to go from this relation to a relation between growth rates—of output, money and the price level. Fortunately, this is easy.
I.et gui be the growth rate ol output, l.et 7Г, be the growth rate of the price level—the rate of inflation. Instead ol having the central bank controlling the level ol the nominal money supply, assume that the central bank has a lixed money growth rule—it keeps lixed the growth rate of nominal money, which we label gml. Then, from equation (9.6), the dynamic AD relation becomes:
S}и = Smt ~ Щ
if nominal money growth exceeds inflation, real money growth is positive and so is output growth. It nominal money growth is less than inflation, real money growth is negative and so is output growth. In other words, given inllation, expansionary monetary policy (high nominal money growth) leads to high output growth; contractionary monetary policy (low nominal money growth) leads to low, possibly negative, output growth.
Now we will show how we can get an analogous dynamic AD relation with an interest rate rule.
A dynamic AD relation when the central bank controls the interest rate
The goods market equilibrium relation in level terms can be expressed as:
Y, = Y(<„C„T,) (-+,-)
(9.9)
This equation simply repeats what you know from Chapters 3 and 5 (lor example, look back at equation [5.2]). An increase in the interest rate reduces investment and thus aggregate demand and output. We are not interested in fiscal policy in this chapter, and so we are going to ignore G, and T(. Merely for convenience, assume now that the goods market equilibrium condition can be reduced to the following simple form:
Y Y — ' i - 11 .
We are simplifying aggregate demand Y(MIP.C.T) in two ways:
• We focus on the
4 relation between the real money stock, MlP. and output. Y. ignoring the two fiscal policy variables. G and T. We can then write the aggregate demand relation as У =Y(MlP).
• We assume that the relation between the real money stock and output is linear.This implies we can write the aggregate demand relation as Y = --MlP.
4 If a variable is the ratio of two variables, its growth rate is equal to the difference between the growth rates of these two variables (see proposition 8 in Appendix 2 at the end of the book). So. if Y = yM!P and у is constant, g, = g„- тг.
U
Dynamic IS relation: ► g*<0=>gfl
This is a non-linear IS relation which still says that output will he lower at higher interest rates. Even il the interest rate doesn't change, output can change because ol Yh which represents all those factors that are associated with output, income and aggregate demand growing at the normal rate, which we defined in the paragraph above equation (9.31. Hy applying proposition 8 in Appendix 2, we can now write this IS relation in a dynamic lorm:
Hut ~ $v Kit
where gvt is the actual growth rate of output, is the normal growth rate i assumed constant > and is the rate of change ot the interest rate. This says simply that, il the interest rate doesn't change, the economy will grow at the normal rate. Il the interest rate is increasing, output growth falls accordingly.
(9.1 I)
Thc interest rate rule we used in equation (7.4» was in terms ol the gap between the price level and the target. P, - P'. In Chapter 7 we wanted to avoid issues to do with inflation- now we want a more realistic description of the central hanks interest rate changes in the context ol a non-zero inflation target. What we wish to capture is the (act that the central bank increases thc rate ol interest whenever the actual inflation rate. ~ rises relative lo its inflation target тт'. In medium-run equilibrium, inflation will be equal to thc target, and so the interest rate will remain unchanged. l or the purposes of this chapter, all we need is the following simple dynamic interest rate rule:
Sit =
Now we have thc ingredients lor our dynamic aggregate demand relation between inflation and output growth. Substitute equation (9.1 1 i into (9.10). This yields:
dUTT, - 7Г7 )
Zyt = S'y
This equation slates that:
Output will grow at less than the normal rate ol growth if inflation exceeds the target rate.
• Growth will be above normal il inflation is below target.
• When inllation is on target, which will be true in the medium run, output will grow at its normal rate.
What is the explanation? Given thc target тт', il inflation goes up. the central bank raises the interest The AD relation in ► rate, which weakens aggregate demand. This means that actual aggregate demand growth and output growth terms: growth will (all relative to the normal rate ol growth. By how much depends on how tough the central bank is on inflation—or how big d) is. Equation (9.12) is a rough simplification that makes our life easier. You should keep in mind that behind this relation hides the mechanisms yoti saw in the IS-LM model and the adjustments ol the nominal rale ol interest by the central bank so that it approaches its medium- run equilibrium value.
(9.13)
As a rough guide. let's see what the Australian data from 1982 to 2007 tell us about equation (9.12). Using the growth rale ol real chained GDP and the inflation rate for the GDP deflator, our regression yields thc following significant estimates:
gyl = .3.2% - 0.45(77, - 7ГГ)
with тт' estimated to be 2.9 per cent from 1001 and 7.4 per cent before that. This tells us that il the gap between inflation and the RBA's target rises by I per cent, output growth will be reduced by 0.45 per cent.
(9.12)
We will use this dynamic ►
AO relation (equation [9.11]) in the rest of the chapter.
7Г'>7ГГ 7г'< T?
■ г* < г,
Compare the forms of the dynamic AD relation in equation (9.12) derived lor an interest rate rule and equation (9.7) derived for a lixed money growth mle. They are similar. From now onwards in this
chaptcr, wc will use the interest rate rule version in equation (9.12), but the conclusions we derive in the next sections will be similar to what wc would get with equation i9.7). In the exercises lor the chapter there are questions you can work on to see the differences when using equation (9.7).
9.2 THE EFFECTS OF A MONETARY POLICY DISINFLATION
Output growth
Okun's law
/
\
Inflation
Phillips curve
Unemployment
Figure 9.2 Output growth, unemployment and inflation
Dynamic aggregate demand
The medium run
Assume that thc centra! bank maintains a constant inflation target, тг1. What will be the values of output growth, unemployment, inllation and the nominal interest rate in the medium am?
• In the medium run, the unemployment rate must be constant,- the unemployment rate cannot be increasing or decreasing forever. Putting u, - мм in Okun's law implies that g./t = gy. In the medium mn, output must grow at its normal rate of growth, gv
• With output growth equal to the dynamic aggregate demand relation implies that inflation is constant and equal to the target rate, it' .
In thc medium run, the central bank must allow nominal money growth to be equal to inflation plus normal output growth: gM = тг7 i g^ This follows from the delinition ol the LM relation given in Chapter 5. The way to think about this result is as follows: since the interest rate will be constant in the medium run, a growing level of output implies a growing level of transactions and thus a growing demand for real money. So, if output is growing at 4 per cent, the real money stock must also grow at 4 per cent per year. If inflation is on target at 2.5 per cent, the central bank must allow the nominal money stock to grow at 4% + 2.5% - 6.5% per year.
• If inllation is constant, then inflation this year is equal to inflation last year: тг, ■ tт,_, = тг1. Looking at the Phillips curve, this implies that u, = u„. In the medium run, the unemployment rate must be equal to the natural rate of unemployment.
If you rccall our discussion in Chapter 7 of fiscal deficits and oil price shocks, the medium-ain interest rate i'„ is affected by permanent shocks to exogenous demand and supply shocks in our closed economy. For example, we saw how a budget delicit reduction implied a lower interest rate in the medium ain. In thc same way, we could easily add liscal policy and oil pricc changes into this chapter, but we leave those for exercises.
The results in this chapter are the natural extension ot the results derived in Chapter 7. There, you saw that changes in the price level target implied lhat money was neutral in the medium run: they had no effect on either output or unemployment, but were reflected one for one in changes in the equilibrium price level and money supply. We see here that a similar neutrality result applies to changes in the target inflation rate, and the implied rate of growth of nominal money: changes in the target and equilibrium inflation rate have no effcct on output or unemployment in the medium run, but are reflected one for one in changes in the rate of money growth
Another way to state this last result is that inflation in the medium run is intimately related to nominal money growth adjusted for normal output growth. Milton Friedman put it this way: Inflation is always and everywhere a monetary phenomenon. Unless the central bank allows higher (adjusted) nominal money growth, lactors such as the monopoly power ot firms, strong unions, strikes, liscal deficits, the price ol oil, and so on cannot have any effect on the inllation rate in the medium run.
The results ot this section are summarised in Figure 9.3, which plots the unemployment rate on the horizontal axis and the inllation rate on the vertical axis.
• In the medium run, the unemployment rate is equal to thc natural rate of unemployment. Thc economy must be somewhere on the vertical line at n = u„.
• In thc medium run, adjusted nominal money growth must be equal to inflation—the rate of nominal money growth minus the normal rate of growth of output equals the target inflation rate. This is represented by the (upper) horizontal line at тг = g,„-gу
A decrease in the target inflation rate Irom ттг to тг'' shifts thc horizontal line downward, moving the equilibrium from point A to point A'. The nominal money growth rate decreases by the same amount as the decrease in thc inflation rate. There is no change in the unemployment rate, which is still equal to u„.
Medium run:g, = g, ►
Medium run: ► ~= 77T: g„ = 7TT + |y
Medium run: и = u- ►
'Unless' is important ► During episodes of very high inflation (Chapter 24), you will see that fiscal deficits often lead to nominal money creation and to higher nominal money growth.
Having looked at where the economy ends up in thc medium ain, we now look at how it gets there. In other words, we look at the dynamics ol adjustment. This is the locus of the next three sections.
Figure 9.3 Inflation and unemployment in the medium run
я " Sm ~ gy
i
я =SM ~gy
Decrease in inflation target from n7 to л7'
Natural
unemployment rate
/
Unemployment rate, u
1 < in > g, >g * ' Now look at the dynamic aggregate demand equation: eventually, inflation is sufficiently low and the interest rate will actually have to go below its new medium-run value, so that output growth can become lor a time greater than the normal rate. We know this must happen because otherwise there is no way to get unemployment back down to the natural rate. g, > g > u . ► • Now look at Okun's law: if output growth is above normal then unemployment starts decreasing.
This gives us a second set ol results. So long as unemployment remains above the natural rate, inllation decreases. Eventually, inflation is sufficiently low, and. by implication, the interest rale will be sufficiently low so that unemployment must start decreasing. Unemployment doesn't stnv high forever.
This is an important pair of results: alter a decrease in the inllation target, unemployment initially increases i the lirst result but eventually turns around t ihe second result >.
Make sure to distinguish between deflation (decrease in the price level) and disinflation (decrease in the inflation rate). ► When should you use 'percentage point" rather than 'per cent'? Suppose you are told that the unemployment rate, which was equal to 10 per cent, has increased by 5 per cent Is it 5 per cent of itself, in which case the unemployment rate is ► (1.05) times 10% = 10.5%! Or is it five percentage points, in which case it is 10% + 5%= 15%? The use of percentage point' rather than 'per cent' helps avoid the ambiguity. If you are told that the unemployment rate has increased by five percentage points, this means that the unemployment rate is 10% + 5%= 15%.
We want to know more, however: For how long does unemployment increase? By how much? How docs it return to the natural rate of unemployment in the medium run? If the central bank cares about unemployment in addition to inflation should it raise the interest rate a lot at once, or should it raise it slowly over time? To answer these questions, we need to look more closely at our three relations.
How much unemployment? And for how long?
We know from the previous sub-section that lower inflation requires initially a higher interest rale. We also know that a higher interest rate implies an increase in unemployment for some time. For the central bank, the question now is: Having decided to aci, at what pace should it proceed? To put it another way, docs it stick to its interest rate rule, or does it use its discretion lo choose a path of the interest rate to get to the medium run quickly?
A first pass
A first pass at the answer can be given by using the Phillips curve relation in equation 9.51:
7Г, - 7T,_, = -a(u, - un)
This relation makes it clear thai disinflation—a decrease in inflation—can be obtained only at the cost ol higher unemployment. For the leli side of the equation lo be negative—thai is, lor inflation to decrease—the term (и, — tt„) must be positive—the unemployment rate must exceed the natural raie
I he equation actually has a stronger implication: the total amount of unemployment required for a given decrease in inflation doesn't depend on the speed at which disinflation is achieved. In other words, disinflation can be achieved quickly, at the cost of very high unemployment lor a few years,- or it can be achieved more slowly, with a smaller increase in unemployment spread over more years. In both cases, the total amount of unemployment, summing over the years, will be the same.
Lei's see why. Define lirst a point-year of excess unemployment as a dilference between the actual and the natural unemployment rate of one percentage point for one year. For example, il the natural rate of unemployment is 6 per cent, an unemployment rate of 8 per cent tor four years in a row corresponds to 4 times (8 - 6} - 8 point-years ol exccss unemployment.
Now look at a central bank thai wants to reduce inflation by x percentage points. To make things simpler, let's use specilic numbers. Assume that the central bank wants to reduce inflation from 7 per cent to 2 per cent, so that x is equal lo 5. Leis also assume thai a equals 0.25 (which is iust below our estimate of 0.26 in this chapter lor the period since 1982).
• Suppose that the central bank wants to achieve the reduction in inflation in one year. Equation (9.5) with a = 0.25 tells us that what is required is one year of unemployment at 5/0.25 20% above the natural rate. In this case, the unemployment gap on the right side of the equation is equal to 20 per cent, and the inflation rate decreases by 5 per cent within a year.
• Suppose thai ihe central bank wants to achieve the reduction in inflation over two years. Equation (9.5) tells us that what is then required is two years of unemployment ai 10 per ceni above the natural rate. During each ol the two years, the unemployment gap on the right side of the equation
is equal to 10 per cent, so the inflation rate decreases hy 2.5 per cent each year, thus by 2 times 2.5% = 5% over two years.
• By the same reasoning, reducing inflation over five years requires five years of unemployment at 4 percent above thc natural rale (5 times 4% = 20%Ь reducing inllation over left years requires len years of unemployment at 2 per cent above the natural rate (10 limes 2% = 20% and so on. Note that in each case the number ol point-years of excess unemployment required to decrease inllation is the same—namely, one year times 20 per cent excess unemployment in the first scenario, two years times 10 percent in the second, and ten years times 2 percent in the lasi. Thc implication is straightforward: the central bank can choose the distribution ol excess unemployment over time, but it cannot change thc total number ol point-years ol excess unemployment.
We can state this conclusion another wav. Define the sacrifice ratio as the number ot point-years of excess unemployment needed to achieve a decrease in inflation of I per cent. Then equation (9.5) implies that this ratio is independent ol policy and simply equal to 1 1/a). We estimated that a equals 0.26 in Australia since 1982, which means the sacrifice ratio is just below 4. To decrease thc inllation target in Australia by one percentage point, the Reserve Bank could force unemployment to be about lour percentage points higher for a year, two points higher lor two years, and so on.
II the sacritice ratio is constant, does this imply that the speed of disinflation is irrelevant^ No. We know that inflation was reduced by live percentage points Irom a value of about 7 per cent in 198') to about 2 per cent in 1994 in Australia. Suppose that the Reserve Bank tried to achieve that live- percentage point decrease in inflation in iust one year. As you have just seen, this would require an unemployment rate of just below 20 per cent above the natural rate for one year. With a natural unemployment rate ol about 6.5 per cent, this would require increasing the actual unemployment rate to 26.5 per cent tor one year, from Okun's law, using a value of 0.5 lor /3 and a normal output growth rale ol .3.3 per cent, output growth would have to satisly
26.5% 6.5% = -0,5(gyl - 3.2%)
This implies a value for - -(20% 1/0.5 + 3.2% -46.8%. In words: Output would have to almost halve for a pear! For comparison, the largest negative growth rate in Australia in the twentieth century was -11.8 per cent in 1915. during the World War I. and the next was -10 per cent in 193 I. during the Great Depression, h is lair to say that macroeconomists don't know with great confidence what would happen il monetary policy were aimed at inducing such a large negative growth rate. But most would surely be unwilling to try. The increase in thc overall unemployment rate would lead to extremely high unemployment rates lor some groups—specifically the young and the unskilled, whose unemployment typically increases more than the average unemployment rate. Not only would the welfare costs lor these groups be large, but such high unemployment might leave permanent scars. The sharp drop in output would most likely also lead to a large number of bankruptcies, with long-lasting effects on economic activity. In short, thc disruptions from a last disinflation are likely to be very large. Ibis suggests that the central bank will want to go more slowly and achieve disinflation over a number of years rather than all in one year.
Working out the dynamics using the model
Our lirst pass discussion ot the dynamics ol a disinflation gave a briel and intuitive explanation of what happens to inflation and unemployment. I lere we take a more detailed look.
For this disinflation experiment, thc central bank wants to decrease the inflation rate Irom 7 percent to 2 per cent. Clearly, it docsn t control either inflation or unemployment directly. Look at figure 9.2 again: what the central bank controls is the interest rate which in turn affects output growth, which in turn affects unemployment, which in turn aflects inflation.
Sacrifice ratio = Point- years of excess unemployment/Decrease
4 in inflation.
4 From equation (9.5), excess unemployment of I per cent for one year decreases the inflation rate by ii times I per cent. Put the othe- way. to reduce the inflation rate by I per cent excess unemployment must be equal to 1/a for one year.
Our three-equation model, equations 9.2) 9.6) and 9.13), assumes that the central bank sticks to its interest rate mlc in equation (9.1 I) lhat we used to derive (9.1 3) and uses no further discretion when selling iis interest rate. But suppose it did noi stick to its rule. Then it could use its discretion to find a path of interest rates over time thai gives a solution lor inflation and unemployment lhat looks like lhat
in Figure 9.4. which is what our first pass discussion suggested. To achieve that first pass outcome, the interest rate will have to be lowered well before year 6 and carefully adjusted thcrealter. There are many paths of the interest rate that could give the same outcome in terms of the number ol point years of excess unemployment but different time paths. All the paths wc would draw share one characteristic: the total unemployment cost—that is, the number ot point-years of excess unemployment—would be the same. Put another way, unemployment hits to remain above the natural rate by a large enough amount, or long enough, to achieve disinflation.
The path displayed in Figure 9.4 was one that might be achieved if the central bank suspended its rule in equation (9.1 I). and used its discretion in setting the interest rate to achieve its desired time path of the point years of excess unemployment. Instead, if we strictly apply the rule, the path looks a little more complicated. Wc leave the discussion of that path to the appendix of this chapter.
The analysis wc have just developed is close to the type of analysis economists at central banks were conducting in the late 1970s with money supply setting. The econometric model they used, as well as most econometric models in use at the time had the property that policy could change the timing, but not the number of point-years, ol excess unemployment. We will call this the traditional approach. The traditional approach was challenged by two separate groups of macroeconomists. The next section presents their arguments and the discussion that followed.
Figure 9.4 A disinflation path
8 ~i 7 6 - 5 - 4
3 -
2- I - 0
YearO
Year I Year 2 Year 3
7
Year 6+
-1
12
10
» Year 4
Year 5
Unemployment rate (%)
Five years of unemployment above the natural rate of unemployment lead to a permanent decrease in inflation.
9.3 EXPECTATIONS, CREDIBILITY AND NOMINAL CONTRACTS
The focus of both groups was the role of expectations, and how changes in expectation formation might affect the unemployment cost of disinllation. But despite this common focus, they reached quite different conclusions.
Expectations and credibility: The Lucas critique
The conclusions of the lirst group were based on the work of Robert Lucas and Thomas Sargent of the University of Chicago. In what has become known as the Lucas critique, Lucas pointed out that when trying to predict the elfccts ot a major change in policy—such as the change considered by the l ed at the time—it could be very misleading lo take as given the relations estimated Irom past daia.
In the case of the Phillips curve taking equation (9.5) as given was equivalent to assuming that wage setters would keep expecting inllation in the future to be ihe same as in the past, and lhal ihe way wage setters lormed expectations wouldn't change in response to the change in policy. This was an unwarranted assumption, i.ucas argued. Why shouldnt wage setters take policy changes into account? Il wage setters believed thai the Fed was committed to lower inflation, they might well expect inflation to be lower in thc future than it had been in the past. It they lowered their expectations ol inflation, then actual inflation would decline without the need lor a protracted recession.
77,
If - тт,the Phillips curve is given by 7Г,-п-и = -а(о,-и„). To achieve 77, < тг,.,. it must be that u, > u„.
i The 'credibility view': fast disinflation is likely to be more credible than slow disinflation. Credib lity decreases the unemployment cost of disinflation.Thus. the central bank should go for fast disinflation.
Fischer is now the ^ governor of the central bank of Israel.Taylor was the US undersecretary for international affairs in the G.W. Bush administration and is now a professor at Stanford University.
The logic ol Lucas's argument can be seen by returning to equation (9.4), the Phillips curve with expected inflation on the right:
-a(u, - u„)
It wage setters kept forming expectations of inflation by looking at last year's inllation (il ттЧ = 7Г,_,), then the only way lo decrease inllation would be ю accept higher unemployment for some time; we explored the implications ol this assumption in the preceding section.
But il wage setters coulc be convinced that inllation was indeed going to he lower than in the past, they would decrease iheir expectations of inflation. This would in turn reduce actual inflation, without there necessarily being any change in lhe unemployment rate. For example, il wage setters were convinced that inflation which had been running at 14 per cent in the past, would he only 4 per cent in the future, and if they formed expectations accordingly, then inflation would decrease to 4 per cent even il unemployment remained at the natural rate of unemployment:
77. = 77',' - a(u, - »„)
4% = 4% - 0%
Interest rales, inllation and expected inflation could all be reduced without the need lor a recession. Put another way, money could be neutral not only in the medium run but also in thc short run.
Lucas and Sargent didn't believe that disinflation could really take place without some increase in unemployment. But Sargent, looking at lhe historical evidence on the end of several very high inflations, concluded that the increase in unemployment could be small. The sacrifice ratio—the amount of excess unemployment needed to achieve disinflation—might be much lower than suggested by the traditional approach. The essential ingredient ot successful disinflation, he argued, was credibility ot monetary policy—the belief by wage setters lhat the central bank was truly committed lo reducing inflation. Only credibility would lead wage setters ю change the way they lormed expectations. Furthermore, he argued, a clear and quick disinflation program was much more likely to be credible lhan a protracted one that offered plenty of opportunities for reversal and political infighting along the way.
This line ot thought has been largely responsible lor many central banks explicitly declaring an inflation target. Thc belief is that so long as they are seen to act to achieve that target, they can enhance thc credibility of their monetary policy, which then leads to a reduction in thc sacrilice ratio.
Nominal rigidities and contracts
A contrary view was taken by Stanley Fischer, then ai Massachusetts Institute ol Technology, and John Taylor, then at Columbia University. Both emphasised the presence ol nominal rigidities, meaning that, in modern economies, many wages and prices arc set in nominal terms lor some time and are typically not readjusted when there is a change in policy.
Fischer argued lhat, even with credibility, loo rapid a tightening ol monetary policy would lead lo higher unemployment Even il the central bank fully convinced workers and firms thai it was intern on lower inflation, the wages set before the change in policy would reflect expectations of inflation prior to thc change in policy. In effcct, inflation would already be built into existing wage agreements, and couldn't be reduced instantaneously and without cost. At the very least, Fischer said, a policy ol disinflation should be announced sufficiently in advance ot its actual implementation to allow wage setters to take it into account when setting wages.
Taylor's argument went one step further. An important characteristic ol wage contracts, he argued, is that they are not all signed at the same time. Instead, they are staggered over time. He showed that this staggering of wage decisions imposed strong limits on how last disinflation could proceed without triggering higher unemployment, even il the central bank's commitment to inflation was fully credible. Why the limits? If workers cared about relative wages—that is, cared about their wages relative to the wages ot other workers—each wage contract would choose a wage not very different from wages in the other contracts in force at the time. Too rapid a monetary policy tightening wouldn't lead to a quick decrease in inllation. Rather, the interest rate would be very high, triggering a recession and an increase in the unemployment rate.
Taking into account the time pattern ot wage contracts in the United Stales. Taylor then showed that, under lull credibility ot monetary policy, there was a path ol disinflation consistent with no increase in unemployment. This path is shown in Figure 9.5
Disinflation starts in quarter I and lasts tor 16 quarters. Once il is achieved, the inflation rate, which started at 10 per cent, is 3 per cent. The striking feature is how slowly disinflation proceeds at the beginning. One year (four quarters) after the announcement of the change in policy, inflation is still 9.9 per cent. But then disinflation proceeds more quickly. By ihe end of the third year inllation is down to 4 percent, and by the end of the fourth year the desired disinflation is achieved.
The reason for the slow decrease in inflation at the beginning—and. behind the scene, lor the slow increase in the interest rale—is straightforward: wages in force at the time ol the policy change are the result ol decisions made betore the policy change, so that the path ol inllation in the near future is largely predetermined. If the interest rate were lo increase sharply, inflation couldn't decrease very much right away, and the result would be a higher interest rate and thus a recession. So. the best policy is lor the central bank to proceed slowly at ihe beginning while announcing it will proceed taster in the luture. This announcement leads new wage agreements to take the new policy into account. When most wage decisions in ihe economy come from decisions made after the change in policy, disinflation can proceed much more quickly. This is what happens in the third year following the policy change.
Like Lucas and Sargent, Taylor didn't believe thai disinflation could really be implemented without an increase in unemployment. For one thing, he realised thai the paih ol disinflation drawn in Figure 9.5 might not be credible. The announcement, say, this year, lhat interest rates will be higher two years from now is likely lo run into a serious credibility problem. Wage sellers are likely to ask themselves: It the decision has been made to disinflate, why should the central bank wait two years? Without
15.0-1
12.5
Figure 9.5 Disinflation path without unemployment in the Taylor model
10.0 -
7.5 -
с 5.0-
2.5 -
1 1 1 1 1 1 1 1 1
0.0
"l 1 Г"
-5 -4 -3 -2
-1 1 1
2 3 4
6 7 Quarters
9 10 II 12 13 14 15 16 17 18
With staggering of wage decisions, disinflation must be phased in slowly to avoid an increase in unemployment
credibility, inflation expectations might not change, defeating thc hope that disinflation can he achieved without an increase in the unemployment rate. But Taylor's analysis had two clear messages. First, like Lucas and Sargent Taylor's analysis emphasised the role of expectations. Second, it suggested that a slow but credible disinflation might have a cost lower than that implied by the traditional approach.
Who turned out to be right: the traditional approach, the Sargent—Lucas approach, or the Fischer- Taylor approach? The answer is given in thc next section, which looks at the Australian disinflation in the early 1990s, and it is easy to summarise. The disinflation ot about 5 per cent triggered a deep recession with about lifteen point-years ot excess unemployment. In other words, there were no obvious credibility gains and the sacrifice ratio turned out to be roughly what was predicted by the traditional approach.
9.4 THE AUSTRALIAN DISINFLATION, 1989-94
In 1989 the Australian unemployment rate was 6.0 per cent CDP growth was 4.5 per cent, and the inflation rate using the CPI index) was considered a high 7.5 per cent. The question the Reserve Bank had been pondering wasn't whether it should reduce inflation but how last it should reduce it. It had already begun the process ol raising the interest rate in mid-1988—it was worrying about what it saw as excessive asset price inflation, which could have been a precursor to an overheated economy.
By December 1989 the cash rate had reached a peak above 18 per cent, the highest rate since Federation. This is shown in Figure 9.6, which plots the cash rate and the inflation rate, measured as the rale ot change ot the CPI over the previous twelve months tor the period beginning in 1988 and ending in 1994
The reason lor lhe decrease in the cash rate over the next three years was that the economy had been forced into a deep recession, with output growth at its worst in 1991. Llneniployment climbed and continued to do so until 1994. Inflation indeed did come down rapidly from 1991, reaching nearly 2 per cent by 1994.
A substantial increase in the interest rate to above 18 per cent in 1989 was followed by a gradual fall over the next five years. SOURCE: RBA bulletin.Tallies Fl and Gl.
The Reserve Bank had demonstrated that it was determined to beat down inflation. It made numerous statements declaring its intention to defeat inflation. By its action ot pushing thc rate ol
December 1989
Cash rate
1—I—I—I—Г
1988
1989
"I—I—I—I—l—I—I—I—I—I—I—l—ri—l—l—l—l—l—I—I—I
1990 1991 1992 1993 1994
Figure 9.6 The cash rate and inflation rate in Australia, 1988-94
interest so high in 1989, it had surely won credibility. Did this credibility gain lead to a more favourable trade-off between disinflation and unemployment than was implied by the traditional approach? Table 9.1 gives the relevant numbers.
The first two lines of the table make it clear that there was no expectation miracle—line 2 shows that disinflation was associated with substantial unemployment. The average unemployment rate was above 9 percent from 1991 to 1994, peaking at 10.8 percent in March 1993. Indeed, it was not until 2000 that unemployment returned to about 6 per cent:
I he answer to whether the unemployment cost was lower than implied by the traditional approach is given in the rest of the table. Under the traditional approach discussed in Section 9.3) each point of disinflation is predicted to require about three point-years of excess unemployment. Line 4 calculates the cumulative number of point-years of excess unemployment from 1990 on, assuming a natural rate of unemployment of 6.3 per cent. Line 5 calculates cumulative disinflation—the decrease in inflation starting from its 1989 level. Line 6 gives the sacrilice ratio, the ratio of the cumulative point-years of unemployment above the natural rate of unemployment to cumulative disinflation.
The table shows there were some 'credibility gains in the beginning. By 1991 the sacrifice ratio looked quite attractive: the cumulative decrease in inflation since 1989 was nearly 4.3 per ccnt. at a cost ol 3.4 point-years ol unemployment— a sacrifice ratio of about 0.8. relative to the sacrifice ratio of above three predicted by the traditional approach. However, the sacrifice ratio showed decreasing gains the longer unemployment remained high, and was nearly back to three by 1994. A 5.6 per ccnt dis¬inflation had been achieved with 15.1 point-years ol excess unemployment an outcome not very different from the outcome predicted at that point by the traditional approach.
In short the Australian disinflation ot the early 1990s was associated with a substantial increase in unemployment. The Phillips curve relation between the change in inllation and the deviation ot the unemployment rate irom the natural rate proved more robust than many economists anticipated.
Was this outcome due to a lack ol additional credibility ol the change in monetary policy, or to the lact that credibility is not enough to substantially reduce the cost of disinflation? One way ot learning more is to look at other disinflation episodes. This is the approach followed by Laurence Ball, Irom Johns Hopkins Llniversity. Ball estimated sacrifice ratios lor sixty-live disinflation episodes in nineteen ORCD countries over thirty years. He reached three main conclusions.
• Disinflations typically lead to a period ol higher unemployment. Put another way, even if monetary policy is neutral in the medium run unemployment increases lor some time before returning to the natural rate of unemployment.
• Faster disinflations arc associated with smaller sacrifice ratios. This conclusion provides some evidence to support the expectation and credibility effects emphasised by Lucas and Sargent.
Table 9.1 Inflation and unemployment in Australia, 1989-94
1989 1990 1991 1992 1993 1994
1. GDP growth (%) 4.5 1.8 -0.7 2.1 3.8 4.8
2. Unemployment rate (%) 6.0 6.8 9.2 10.5 10.7 9.4
3. CPI inflation (%) 7.5 7.3 3.2 1.0 1.8 1.9
4. Cumulative unemployment (%) 0.5 3.4 7.6 12.0 15.1
5. Cumulative disinflation (%) 0.3 4.3 6.6 5.7 5.7
6. Sacrifice ratio 1.8 0.8 1.2 2.1 2.7
Cumulative unemployment is the sum of point-years of excess unemployment from 1990 on, assuming a natural rate of unemployment of 6.3 per cent Cumulative disinflation is the difference betv/een inflation in a given year and inflation in 1989. The sacrifice ratio is the ratio of cumulative unemployment to cumulative disinflation. SOURCE: RBA BAtin.Tables G10. G7. FI.
• Sacrifice ratios are smaller in countries that have shorter wage contracts. This provides some evidence to support Fischer and Taylor's emphasis on the structure of wage agreements.
Let's summarise. Policy-makers face a trade-off between unemployment and inflation. In particular, to permanently lower inllation requires higher unemployment lor some time. One might have hoped that, with credible policies, the trade-oil would he much more favourable. Thc evidence can be read as saying that credibility gains may be present but they arc small.
SUMMARY
• There are three relations linking output, unemployment and inllation.
Okun's law shows how the deviation ol output growth from normal leads to a change in the unemployment rate In Australia today, based on data since 1982 output growth of I percent above the normal rale (which is 3.2 per cent) for a year leads to a decrease in the unemployment rate of about 0.5 per cent.
The Phillips curve shows how thc deviation of the unemployment rale from the natural rate leads to a change in the inllation rate. In Australia today, based on data since 1982, an unemployment rate 1 per cent below the natural rate for a year leads to an increase in the inflation rate ol 0.26 per cent.
The dynamic aggregate demand relation shows how the inflation affects output growth. With an interest rate rule higher inllation relative to the inflation target leads the central bank to raise the interest rate, causing a decrease in output growth relative to the normal rate ol growth. In Australia today, based on data since 1982, an inflation rate I percent above the central banks target leads to a decrease in output growth by 0.45 per cent.
• In the medium ain. the unemployment rate is equal to the natural rate of unemployment, and output grows at its normal growth rate.
• In the short am, a decrease in the central bank's inllation target leads to a slowdown in growth and an increase in unemployment for some time. Thus disinflation (a decrease in the inllation rate) can be achieved only at the cost ot more unemployment. How much unemployment is a controversial issue.
• Thc traditional approach assumes that people don t change the way they form expectations when monetary policy changes, so that the relation between inllation and unemployment is unallected by the change in policy. This approach implies that disinflation can he achieved by a short but large increase in unemployment, or by a longer and smaller increase in unemployment. But policy cannot affect the total number ot point-years of excess unemployment. If the central bank sticks rigidly to its monetary policy rule unemployment and inllation typically converge in a spiral to the medium- run equilibrium.
• An alternative view is that, il the change in monetary policy is credible, expectation formation may change, leading to a smaller increase in unemployment than predicted by the traditional approach. In its extreme form, this alternative view implies that, if policy is fully credible, it can achieve dis¬inflation with no increase in unemployment. A less extreme form recognises that, while expectation formation may change the presence ol nominal rigidities is likely to imply some increase in unemployment, although less than implied by the traditional approach.
• The Australian disinflation ol the early 1990s during which inllation decreased by approximately 5 per cent, was associated with a large recession. Thc unemployment cost was close to the predictions ol the traditional approach.
Комментариев нет:
Отправить комментарий