UNIVERSITY OF CALIFORNIA Economics 134 DEPARTMENT OF ECONOMICS Spring 2018 Professor David Romer SUGGESTED ANSWERS TO PROBLEM SET 3 1. Automatic stabilizers a. Expenditure curve becomes flatter Government expenditure G is now a decreasing function of Y. As Y increases, consumption increases as before, but G decreases. Thus, the planned expenditure curve becomes flatter. (The figure is drawn under the assumption that it does not become downward sloping, though there is nothing in the problem that rules that out.). Note that E 1 intersects the 45 line at the initial equilibrium, since G(Y ) = G. The IS curve becomes steeper A flatter planned expenditure curve implies a steeper IS curve. You can see this graphically in the Keynesian Cross: A change in the interest rate shifts E 0 and E 1 vertically by the same amount, since only investment changes. However, because E 1 is flatter, the resulting change in output Y is smaller (try working through it yourself). As before, the new IS curve intersects the MP curve at the initial equilibrium, since government expenditure is unchanged at Y. Intuitively, the new government policy makes output less responsive to changes in the real interest rate. When the real interest rate falls, investment and consumption increase as before, raising output. But now, the government counteracts some of the increase in output by reducing its expenditure. Thus, output does not rise by as much as it did before. 1
b. The AD curve becomes steeper A steeper IS curve implies a steeper AD curve. Remember that the AD curve is derived by changing inflation, which shifts the MP curve and tracing out all the resulting equilibria in the IS-MP diagram (This is shown in blue in the diagram). Since the FedED s interest rate rule has not changed, the same change in inflation from π 0 to π shifts the MM curve vertically by the same amount as before. However, since the new IS curve IS 1 is steeper than IS 0, the resulting change in output is smaller (Y 1 < Y ). Thus, the AD curve is steeper. As above, the new AD curve intersects the IA 0 at the initial equilibrium, since government expenditure is unchanged at Y. Intuitively, when inflation increases, the central bank increases interest rates to fight rising inflation. This reduces investment and consumption as before. But now, the government increases its expenditure G to counteract some of the fall in output induced by the central bank raising the real interest rate. Thus, output becomes less reactive to changes in inflation compared to before. c. Are automatic stabilizers desirable? Policies with the effects described above are often called automatic stabilizers, since they reduce output when it is high, and increase output when it is low. Examples include unemployment benefits and other income assistance programs. Automatic stabilizers do two things: First, they automatically stabilize the economy. Suppose a negative shock hits the economy (e.g. an increase in inflation due to an oil price hike, a decrease in inflation expectations, a decrease in consumer confidence, a mistake by the FedED that increased interest rates by too much). With automatic stabilizers, output will fall by less than it would have otherwise, even if the FedED s monetary policy and discretionary fiscal policy are unchanged. Since economic agents are risk averse, they do not like volatility, and automatic stabilizers are therefore desirable from this point of view. Second, they reduce the effectiveness of discretionary monetary policy. If the FedED wants to tighten/loosen its monetary policy stance, this shifts the MP curve up/down. We can see in the diagrams that this lowers/raises output by less than it did without automatic stabilizers. The reason is that fiscal policy automatically leans against monetary policy. Thus, automatic stabilizers are less desirable from a monetary policy makers point of view. 2
There is therefore a trade-off: Which of those arguments is more important depends on whether you think monetary policy can always perfectly stabilize the economy. To the extent it cannot (e.g. because it does not perfectly anticipate all the shocks hitting the economy), automatic stabilizers are useful. Most economists view them as desirable. 2. The Effect of the 2008 Economic Stimulus Payments on Car Purchases This question was motivated by Figure 1 from Valerie Ramey. 1 Using calculations reported in Sahm, Shapiro & Slemrod (2012), 2 she illustrates what the numbers from Parker et al. imply for aggregate expenditures on new cars. Parker et al. estimate that households spent 35.7% of the stimulus payment on new cars within three months of when it was received (Table 7, Panel E). If we interpret this number as the causal effect on aggregate spending on new cars, aggregate spending on new cars within three months of when the payments went out should have been 0.357 * total stimulus payments higher than in the absence of the stimulus. The green line shows actual expenditures on new cars, while the red line shows the implied expenditure in the absence of the stimulus payments. 3. First, we see that aggregate spending on cars did not jump up by the large amount implied by the estimates from Parker et al. in the three months after the stimulus payments were sent out. Second, if Parker et al. s estimates are interpreted as aggregate effects, spending on new cars would have collapsed to almost zero in the absence of the stimulus package and jumped back up to almost its previous level a few months later. This is clearly implausible. Thus, it is likely that the aggregate effect on car expenditures were in fact much smaller than the estimates from cross-sectional evidence by Parker et al. 1 Valerie A. Ramey, "Discussion of Fiscal Policy by Alan Auerbach," November 2017. 2 Sahm, Claudia R., Matthew D. Shapiro, and Joel Slemrod. "Check in the Mail or More in the Paycheck: Does the Effectiveness of Fiscal Stimulus Depend on How It Is Delivered?" American Economic Journal: Economic Policy 4, no. 3 (2012): 216-50. 3 This line is calculated by subtracting 0.357 * total stimulus payments from actual car sales. It assumes, therefore, that in the absence of the stimulus, aggregate spending would have been lower by that amount. 3
Why could this be the case? This question asked you to engage critically with the evidence presented in the Parker et al. (2013) paper. The key estimate is found in Table 7, Panel E. Statistical Significance: The first point to note is that the 95% confidence interval is large. It is calculated as: 95% CC = b 2ss b, b + 2ss b = [ 0.051, 0.765] The confidence interval therefore includes much smaller values than 0.357, and we cannot even reject the null hypothesis that there was no increase in new vehicle purchases at the 5% confidence level. Thus, the first reason the effect might have been much smaller is that it is imprecisely estimated: for purely statistical reasons, we could have estimated such a high coefficient even if the true effect was much smaller. Local Effect vs. Aggregate Effect: The empirical strategy of Parker et al. exploits the fact that the timing of payments was determined by the last two digits of recipients Social Security Number, and therefore essentially random. Their estimate of 0.357 therefore comes from comparing the spending of households that have already received the transfer to those which have not yet received the payment. Given that the randomization is compelling, this is likely a causal effect on the difference in spending between those two groups. However, even if the point estimate of 0.357 corresponds to the truth, it may not correspond to the change in aggregate spending. Aggregate spending on new cars at any time t is given by: C aa cccc (t) = C EEE cccc (t) + C nnnn cccc (t) where C EEE cccc (t) is spending by households that have already received the stimulus payment, and C nnnn cccc (t) is spending by households that have not (yet) received the payment. Thus, the effect of the stimulus on aggregate spending is equal to 0.357 only if households that have not (yet) received the payment spent the same amount on new cars as they would have in the absence of any stimulus payments. Consider the two scenarios in the figure below. Assume that if there had been no stimulus payments, new car purchases C of every household would have been growing at the same rate as before. In the left panel, car purchases of households who have not (yet) received the payment grow at the same rate as they would have in the absence of any stimulus payments. Aggregate spending therefore increases by 0.357 of the total stimulus payments. In the right 4
panel, however, car purchases households who have not (yet) received the stimulus grow more slowly than they would have in the absence of any stimulus payments. While the difference between the two groups remains at 0.357 (as per Parker et al. s results), the aggregate effect is much smaller. Aggregate implications of cross-sectional estimates Why might new car purchases of households who had not (yet) received the stimulus payments been growing more slowly than they would have in the absence of the stimulus? Below is a (nonexhaustive) list: i. Anticipation of future payments: Perhaps, households that had not yet received the payment were anticipating that they would receive it, and delayed their car purchase to a later date when they would get it. ii. Price Effects: Perhaps, the increase in the demand for cars from recipient households increased the price for new cars temporarily. Because of this, households who had not (yet) received the payments would be less willing to buy a new car than they would have been in the absence of the stimulus. iii. Supply-side capacity constraints: Perhaps, there are constraints on the number of cars that suppliers can sell in any given period. There might be physical capacity constraints as the number of cars that can be produced at existing factories is limited, and new factories take some time to build, or there might be constraints on dealers ability to get cars delivered to their lots quickly. iv. Sales Quotas: Car sellers might be working on a quota system where they need to meet a certain target number of cars sold in order to get paid. As soon as they reach it, they might reduce their effort or take some days off, thus holding the number of cars sold constant. 5
3. TRUE/FALSE/UNCERTAIN a. Expansionary fiscal policy failed to end the Great Depression in the 1930s not because it does not work, but because it was not tried. FALSE. This statement is taken from an influential paper (Brown, 1956) 4 and was, for a time, the received wisdom among economic historians. Some of the reasons behind this view were presented in Lecture 14 (see lecture slides 4-16): The fiscal expansion during the New Deal period as measured by the High-Employment Surplus was moderate, and small relative to the size of the depression. Moreover, New Deal fiscal policy varied between expansionary and contractionary. Formatted: Indent: Left: 0.05", Keep with next, Keep lines together Formatted: Indent: Left: 0.3", Keep with next, Keep lines together, Don't adjust space between Latin and Asian text However, Hausman (2016) shows that this statement is incomplete, and that Brown underestimated the effect of the Veterans Bonus. Hausman uses a difference-in-differences approach to show that the bonus led to a large increase in veterans spending relative to nonveterans. 70 percent of the bonus had been spent within 6 months, mostly on cars. Since the bonus was 2.1 percent of 1936 US GDP, Hausman estimates it contributed about 1.6 percentage points to 1936 GDP growth. (Underlying this estimate is an assumption that the fiscal multiplier is equal to 1.) Thus, fiscal policy was tried and did contribute to the recovery from the Great Depression. Note: You could also say that this statement is TRUE, and argue that the veterans bonus was not explicitly designed as fiscal stimulus, or that it came too late, or that it was too small. What is important, is that you engage critically with the evidence in Hausman (2016). b. If the fiscal policy multiplier is less than 1, then fiscal policy cannot help to end a recession. FALSE. Fiscal policy can help to end a recession as long as the fiscal policy multiplier is positive. If the fiscal policy multiplier is greater than zero, then a fiscal stimulus can increase output. Note that this is true even if the fiscal policy multiplier is greater than zero, but less than one. (Actually, if the fiscal policy multiplier is negative, fiscal policy can still be used to end a recession but in that case, the recommended policy would be a fiscal contraction rather than a fiscal expansion!) 4 Brown, E. Cary. "Fiscal Policy in the 'Thirties: A Reappraisal." The American Economic Review 46, no. 5 (1956): 857-79. 6
4. Banking Crises in Ireland and Iceland a) Króna depreciation: It is not obvious which data source one should use for the exchange rate of the króna in these turbulent times. When the banking panic intensified, the Icelandic central bank tried to peg the króna unsuccessfully. In October 2008, trading in the currency collapsed when the last major Icelandic bank was taken into receivership, and thus all króna trade 'clearing houses' were lost. Subsequently, the government introduced capital controls and quotas on how much foreign currency each Icelandic institution was allowed to buy. These measures created a situation where the exchange rate quoted by the central bank wasn t really a market clearing price, and the króna could have been traded at different rates in offshore foreign exchange markets. Here we are going to use the ECB reference exchange rate for the crisis period which is different from the official exchange rate series on the Icelandic central bank s website, but still less extreme than what some news reports from the crisis period suggest. For the value of the ISK in 2013, when markets had normalized, we are going to rely on Bloomberg for a representative market exchange rate. Figure 2: ECB reference exchange rate (ISK/EUR) 5 Figure 2 plots the ECB reference exchange rate (ISK/EUR) between January 2008 and December 2008 when it is available. We can see the sharp depreciation after September 2008. (The exchange rate is reported as króna per euro, so a rise corresponds to a depreciation.) The low point was 305 ISK/EUR which means a depreciation of 92% (computed as 100 times the change in the natural log of the exchange rate) compared to the average August 2008 exchange rate. However, when market conditions normalized, the currency regained most of this initial loss. The króna traded at 157.64 against the euro on 2 August, 2013, only 25.6% below its precrisis level (Table 1). 5 The data come from the ECB and can be found at: https://www.ecb.europa.eu/stats/policy_and_exchange_rates/euro_reference_exchange_rates/html/eurofxrefgraph-isk.en.html 7
b) Unemployment in Ireland vs. Iceland: Table 1 shows the evolution of unemployment in Iceland and Ireland. The unemployment rate increased much more in Ireland, starting from an already higher level. Moreover, Iceland regained more than half of the employment loss by 2013. In contrast, in 2013 Irish unemployment was only slightly below its peak level. Table 1: Pre- and post-crisis comparison for Iceland and Ireland (Eurostat 6 ) Pre-crisis Low point Change Value in 2013 (August 2008) Króna/euro 122.07 Unemployment Iceland 3.4% (seasonally adjusted) Unemployment Ireland 7.5% (seasonally adjusted) * Calculated as the change in natural logarithms. 305 (October 9, 2008) 7.7% (October 2010) 16.0% (January 2012) 92% * 157.64 7 (August 2, 2013) 4.3 ppt 8.5 ppt 5.5% (August 2013) 13.3% (August 2013) c) Net Exports: Figure 3 shows net exports as a share of GDP in the two countries before and after the banking crisis. The trade balance improved in both countries, but the improvement is faster and more pronounced in Iceland, which is consistent with the hypothesis that exchange rate depreciation helped cushion the effects of the banking crisis in Iceland. Figure 3: Net exports as a share of GDP (Eurostat, seasonally adjusted 8 ) % 30 20 Banking crisis (2008Q3) 10 0-10 -20-30 2005Q1 2005Q2 2005Q3 2005Q4 2006Q1 2006Q2 2006Q3 2006Q4 2007Q1 2007Q2 2007Q3 6 The data are seasonally adjusted unemployment rates from Eurostat, series une_rt_m from their database, and can be found at: http://ec.europa.eu/eurostat/web/lfs/data/database 7 The data point is from Bloomberg, corresponding to a representative market exchange rate, and can be found at: https://www.bloomberg.com/quote/eurisk:cur 8 The data are seasonally adjusted net exports from Eurostat, series B11 from their namq_10_gdp database, and can be found at: http://ec.europa.eu/eurostat/web/national-accounts/data/database 8 2007Q4 Ireland 2008Q1 2008Q2 2008Q3 Iceland 2008Q4 2009Q1 2009Q2 2009Q3 2009Q4 2010Q1 2010Q2 2010Q3 2010Q4
Figure 4 gives some more details where the improvement in net exports is coming from. It shows the evolution of real imports and exports in the two economies. Iceland s exporters performed better than exporters in Ireland, which suggests that currency depreciation improved the competitiveness of Icelandic producers. At the same time, imports in Iceland fell remarkably, which suggests the depreciation made foreign goods much more expensive. 9 Figure 4: Export and import volume indexes (2005=100, Eurostat, seasonally adjusted 10 ) 160 140 120 100 80 60 5. C 6. B 7. C 40 20 0 2005Q1 2005Q2 2005Q3 2005Q4 2006Q1 2006Q2 2006Q3 Banking crisis (2008Q3) 2006Q4 2007Q1 2007Q2 2007Q3 2007Q4 2008Q1 2008Q2 Ireland_exp Iceland_exp Ireland_imp Iceland_imp 2008Q3 2008Q4 2009Q1 2009Q2 2009Q3 2009Q4 2010Q1 2010Q2 2010Q3 2010Q4 9 Of course, the fall in imports also reflects the general decline in aggregate demand, since a fraction of consumption, investment and government purchases comes from abroad. However, the improvement of the trade balance (Figure 31) suggests that the proportional decline in imports was much bigger than the decline in GDP. This disproportional decline in imports can be explained by the exchange rate depreciation. 10 The data are seasonally adjusted imports and exports from Eurostat, series P7 and P6 from their namq_10_gdp database, and can be found at: http://ec.europa.eu/eurostat/web/national-accounts/data/database 9