Economist's View: Should We Pop Bubbles?

Should We Pop Bubbles?

June 30, 2009

This may help Brad DeLong settle his inner conflict over whether Greenspan made an error by not moving interest rates to limit the housing boom. Guillermo Calvo and Rudy Loo-Kung argue that the benefits of bubbles almost always outweigh their costs (and thus there's no need for regulation to prevent them).

I think the authors are correct to point out that distributional issues are omitted from the analysis. Also, the assumption that social welfare depends only upon consumption is important as it rules out any utility costs associated with losing a home, a job, changing schools, etc. over and above the loss of consumption. In addition, using the aggregate consumption level of a composite commodity to index social welfare doesn't capture the costs associated with producing the subotimal mix of goods (e.g. too much housing, not enough of other goods), all that matters is the total quantity that is produced and consumed. Finally, I was surprised that the downturn and upturn phases of the cycle were assumed to be of equal length as I thought a slower return to normal growth (as compared to the downturn) - something that would increase the costs of the collapse - was the normal scenario:

Should we rush to further regulate financial institutions?, by Guillermo Calvo and Rudy Loo-Kung, Vox EU:

‘Tis better to have loved and lost,
Than never to have loved at all.
Tennyson, 1850.

In times of systemic financial distress, hunting for culprits becomes a popular sport. The Madoffs of this world are easy targets because crisis makes crookery harder to conceal. While there is no question that crooks should be sent to jail, increasing financial regulation is a different issue and requires careful analysis. Rushing to impose tighter regulations may hamper recovery and growth. Empirical evidence strongly supports the view that growth and financial development go hand in hand (Demirgüç-Kunt and Levine 2008). Although it is much harder to establish that financial development causes growth, few would doubt that, at least temporarily, financial deregulation could promote higher growth. A genuine concern, however, is that the financial sector is prone to crises, which are typically associated with serious effects on output and employment.

We cannot reach definite conclusions about the desirability of risky financial arrangements in a short column. Our objective is much more modest. We examine the welfare implications of financial deregulations that result in higher growth but end in tears and perform the exercise in the context of a benchmark case in which consumption is the ultimate source of welfare, ignoring possibly relevant behavioural finance and political economy considerations. We base our analysis on estimates of the costs of financial crises in emerging market economies (since the 1980s), a cauldron of financial crises in the last thirty years. Our results support deregulation even under those dire circumstances.¹

A model of growth, collapse, and welfare

More specifically, suppose that financial deregulation is implemented at time 0 and that, as a result, consumption grows at rate g^H (where H stands for “high”); after T periods, there is a crisis that produces a (symmetric) collapse-recovery recession phase in consumption, resembling those observed in the 1990s’ Emerging Economies crises (see Figure 1) . That is, we assume that, starting at time T consumption decreases for a while and then begins to recover. The recession phase takes DT periods. During the first half of this phase, i.e., for DT/2 periods after time T, consumption declines at the rate g*; and then, for the next DT/2 periods, consumption resumes growth at the same rate g*. By construction, at time T + DT (end of the recession phase) consumption reaches its pre-crisis level (i.e., the level prevailing at time T). Afterwards, we assume that consumption grows at a lower rate g^L (where L stands for “low”). We assume that g^L is also the growth rate that would prevail if no financial deregulation had been implemented. Thus, this corresponds to a financial deregulation experiment in which when crisis hits authorities get cold feet and meekly go back to the old, low-growth, financial system forever. This extremely pessimistic scenario will allow us to make a stronger case for deregulation.

Figure 1. Consumption paths under alternative regimes for the average emerging economy

Note: The consumption path associated with financial innovation shows the calibrated collapse-recovery phase for the average emerging economy and the calculated break-even T using a degree of risk aversion (σ) equal to 4.

To calibrate DT and g*, we focus on average output collapse and recovery patterns (the recession phase) observed in emerging markets during times of systemic financial turmoil throughout the period 1980-2004, discussed in Calvo, Izquierdo and Talvi (2006).² More specifically, we set DT equal to the time that it took for average output to recover its pre-crisis level. The growth rate g* is calibrated to match accumulated output loss, which is defined as the sum of the differences between the pre-crisis peak GDP and observed GDP within the recession phase. This procedure suggests setting g* = 3.11% per year and DT = 3.43 years.

Moreover, we set g^H equal to the average GDP growth rate observed in emerging markets during 1992-97, a period in which many countries opened up to capital inflows. The low growth rate g^L is set equal to the average growth rate observed in the previous ten years (1982-91). This leads us to set g^H = 4.7% and g^L = 2.7% per year.³

We focus on the following question: How long should the bonanza or high-growth period T last for financial deregulation to be socially desirable? To answer that question, we examine the benchmark case in which welfare can be expressed as the present discounted value of a utility index which depends on aggregate consumption.⁴

We define the break-even T as the number of bonanza years that would make deregulation welfare equivalent to not deregulating at all and generating low growth, g^L, at all times. If the bonanza period exceeds break-even T, then financial deregulation is preferable to doing nothing, even though it results in a painful crisis. Table 1 and Figure 1 summarise the results (parameter σ is the coefficient of relative risk aversion).⁵

Table 1

Emerging market episodes lasted 5 to 6 years on average, implying that the experiments were socially beneficial despite ending in large recessions. Admittedly, the boom-bust episodes are not identical across economies. To test for robustness, we perform the same exercise for two polar episodes in Latin American, namely, Argentina’s and Chile’s, for which the bonanza period was 4 and 13 years, respectively.⁶ In both cases, results point in favour of financial liberalisation for σ = 4. However, in the case of Argentina (and σ=1), the methodology yields borderline results (Chile passes the test with flying colours).⁷

Two points are worth making: (1) support for deregulation is stronger if the coefficient of relative risk aversion is more realistically set at 4, and (2) break-even T is the same if one assumes that the cycle is repeated as many times as desired (high growth-bust-high growth), and only after the last cycle the economy resumes low growth.⁸ This is more realistic because emerging markets returned to exhibiting high growth during 2003-2007.

The analysis abstracts from the important issues of poverty and income distribution, which might alter our assessment of past deregulation episodes, but that does not make our analysis less relevant looking forward. For example, for the type of social welfare function considered here, if income distribution remains fairly constant, one would reach the same pro-deregulation conclusions even if one entirely focused on the welfare of the poor, à la Rawls. This shows that financial deregulation would be desirable under the Rawlsian criterion if one can find suitable social protection mechanisms, and that the effectiveness of those mechanisms should be explored as part of the grand design of new financial regulations – especially before enacting new regulations that would stifle the dynamism of the financial sector.

Conclusion

Our analysis in this column may help explain why policymakers are hesitant to prick the bubble when it starts – they may simply be trying to maximise social welfare and realise that a potential crisis is not strong enough reason to prevent the bubble from developing (Tennyson’s verses ringing in their ears?). Of course, no policymaker likes crises. When crises strike, much of the discussion focuses on how to avoid them or lessen their impact in the future. This is quite understandable. However, this does not insure that “they are not going to fall in love again.” Therefore, the policy debate should give equal time to discussing what to do when crises happen and to developing institutions that help to assuage their blow.

In closing, we would like to point out that even though this note gives some support to financial deregulation, it does not rule out the existence of financial arrangements that are far superior to the ones currently available. A case in point would be the creation of a global lender of last resort. Central banks have successfully filled that role at the local level and likely prevented many serious self-fulfilling banking crises in the last seventy years. However, there is no equivalent to a lender of last resort at the global level. Its absence was clearly felt in emerging markets in the aftermath of the Russian August 1998 crisis. Even the subprime crisis suffered from the absence of a fully effective lender of last resort. To be sure, central banks stepped up to the plate early on in the current episode, but their coverage was and still is quite limited. Many central financial institutions were left without a safety net, or the net was stretched out after they hit the ground. We feel that the issue of a global lender of last resort should be given more weight in the current debate (see Calvo 2009).

References

Baldacci, Emanuele , Luiz de Mello, and Gabriela Inchauste (2002) "Financial Crises, Poverty, and Income Distribution" IMF Working Paper 02/4.
Barro, Robert (2006) “Rare disaster and Asset Markets in the Twentieth Century”, Quarterly Journal of Economics, 121(3).
Calvo, Guillermo (2009) “Lender of Last Resort: Put it on the agenda!”, VoxEU column, 23 March
Calvo, Guillermo, Alejandro Izquierdo and Ernesto Talvi (2006) “Phoenix Miracles in Emerging Markets: Recovering Without Credit from Systemic Financial Crises,” National Bureau of Economic Research, Working Paper 1201, March.
Demirgüç-Kunt, Asli and Ross Levine (2008), “Finance, Financial Sector Policies, and Long-Run Growth,” Commission on Growth and Development, Working Paper No. 11, World Bank, Washington, DC
Rancière, Romain, Aaron Tornell and Frank Westermann (2008) “Systemic Crises and Growth,” Quarterly Journal of Economics, pp. 359-406.

[1] Our results, thus, give further support to the line of research advanced by Aaron Tornell and Frank Westermann since 2002, which is inspired by the conjecture that financial liberalisation may be socially desirable despite the booms and busts that it may generate. See Rancière, Tornell and Westermann (2008) and their recent VoxEU column.
[2] The paper focuses on episodes in which GDP peak-to-trough contraction is greater than the median fall in the sample. Note that including only the most severe collapses in the calibration constitutes a more difficult test for the case of financial deregulation.
[3] Countries included are those tracked by the J.P. Morgan’s EMBI Global Index: Argentina, Belize, Brazil, Bulgaria, Chile, China, Colombia, Côte d'Ivoire, Dominican Republic, Ecuador, Egypt, El Salvador, Gabon, Ghana, Hungary, Indonesia, Iraq, Jamaica, Lebanon, Malaysia, Mexico, Morocco, Pakistan, Panama, Peru, Philippines, Poland, Romania, South Africa, Sri Lanka, Thailand, Trinidad and Tobago, Tunisia, Turkey, Uruguay, Venezuela, and Vietnam.
[4] More concretely, we assume that the utility index exhibits constant relative risk aversion, σ, and the instantaneous rate of discount equals 3% per year.
[5] If parameters are calibrated on the basis of GDP per capita (instead of its level) yields similar results, due to the high correlation between the two series.
[6] In both cases, we set g^L to average GDP growth rates during 1951-1970. The parameter g^H is set to the average GDP growth rates during 1991-94 for Argentina and 1984-97 in the case of Chile; The values and DT and g* are calibrated to match the characteristics of the Argentine crisis of 2002 and the Chilean crisis of 1998.
[7] In an exercise in which the collapse in growth is modeled as a stochastic event with constant probability, following Barro (2006), we also find support for financial deregulation. In both cases, the break-even expected frequency of these events is lower than the ones observed in the data
[8] It follows that T will be the same if the cycle is repeated an infinite number of times.
[9] The empirical work of Baldacci, de Mello, and Inchauste (2002) suggests that the financial crises that struck developing countries between 1960 and 1998 had severe effects on poverty and, in some cases, income inequality.

Posted by Mark Thoma on Tuesday, June 30, 2009 at 12:19 AM in Economics, Financial System, Regulation Permalink TrackBack (0) Comments (37)

Economist's View: FRBSF: Fighting Downturns with Fiscal Policy

FRBSF: Fighting Downturns with Fiscal Policy

Sylvain Leduc of the San Francisco Fed reviews several studies on the effectiveness of fiscal policy and concludes:

The findings from the three empirical studies, particularly those of Romer and Romer and Mountford and Uhlig, suggest that the fiscal stimulus package will boost growth substantially over the next two years, partly because it includes sizeable tax cuts that can be implemented quickly and that have significant effects on output. Nevertheless, the uncertainty regarding those estimates remains high. ...

This brings up a point about tax cuts I've been meaning to make (again). The effectiveness of tax cuts depends, in part, on how hard the recession hits household balance sheets. In a recession where balance sheets are relatively unaffected, a tax cut may very well translate into spending, and do so fairly quickly.

But when balance sheets are hit hard, the result is different. In this case tax cuts may be used largely to rebuild balance sheets - to recover what was lost - rather than for new spending. Thus, in this recession the stimulative effects of tax cuts may not have as large of an immediate effect as in the past (there are also reasons to suspect the government spending multipliers shown in the table below are underestimated due to the fact that this recession is not like those in the data used to produce the estimates, e.g. for one, the historical data may overestimate the crowding out effect, but for now I want to focus on taxes).

The fact that in this recession tax cuts may not have as large of an immediate impact as in the past should not necessarily lead us to conclude that the tax cuts were a waste. That is, households will not turn back to consumption until they have saved enough to make up for what has been lost, at least in part, so how long it takes for the recession to end depends upon how quickly household balance sheets are refilled (once this is over, I expect saving rates to be higher than in the past, but I also expect that saving will fall some from where they are now once balance sheets are in better shape). The faster they are refilled, the sooner people begin to spend more, and the sooner this thing ends.

So in that sense, the tax cuts were not a waste at all. Unfortunately, however, during the time when the balance sheets are being refilled it will look like the stimulus package is not having any effect - all you see is higher savings rate - but again, the higher saving rate brings the end of the recession nearer in time, and that is important in and of itself. That's a hard effect to estimate, even if you are looking for it after the fact, but again, it shouldn't be dismissed as inconsequential.

Finally, because tax cuts are likely to be saved more than in the past, and hence have a smaller impact than tax multipliers from historical data suggest, and because there is reason to think that government spending multipliers rise as recessions get more severe (e.g. even if interest rates go up, investment is likely to be insensitive when conditions are bad), the logic of using both tax cuts and spending to stimulate the economy is sound.

Here's more:

Fighting Downturns with Fiscal Policy, by Sylvain Leduc, FRBSF Economic Letter: Should fiscal policy be used to fight recessions? Most economists would answer that, for normal economic ups and downs, business cycle stabilization should be left to monetary policy and that fiscal policy should focus on long-term goals. The main argument is that monetary policy can act quickly when output falls below an economy's potential or when inflation varies from its optimal rate, and that these actions can be reversed quickly as conditions change. By contrast, modifications to the fiscal code take a long time to enact and implement and can be very difficult to undo.

However, the current recession is clearly not a typical downturn. In particular, unlike other post-World War II U.S. recessions, monetary policy has run out of its usual ammunition to boost economic activity. The federal funds rate, the principal tool that the Federal Reserve uses to stabilize the economy, is now hovering near zero. Because interest rates cannot be negative in nominal terms, monetary policymakers are unable to lower the federal funds rate further. In this situation, the Federal Reserve has turned to unconventional tools to get around this barrier, commonly called the zero lower bound.

Because of the severity of the recession and the uncertain effects of unconventional monetary policy tools, Congress and the Obama Administration have also enacted a fiscal stimulus package. The $787 billion program approved by Congress in February includes a mix of tax and spending measures aimed at creating jobs and boosting output. Yet, economists and political leaders heatedly debate whether tax cuts or increased spending are more effective, a dispute that's hard to resolve because of the difficulty of determining the precise magnitude of fiscal policy's impact on real GDP. This Economic Letter examines some recent empirical studies analyzing data on the relative effects of higher spending and lower taxes on output.

A simple theory of the effects of fiscal policy

Basic Keynesian theory suggests that the effect of a change in fiscal policy on real GDP is more than one-for-one. For instance, since government spending is one component of GDP, an increase in government purchases, by putting idle resources to work, boosts income one-for-one when the money is initially spent. In addition to that, though, since consumption is a function of current after-tax income in this framework, households also increase their consumption in line with their higher incomes, multiplying the effect of the initial government spending on GDP. The "multiplier effect" of government spending on GDP is thus greater than one.

This simple framework also predicts that the multiplier effect of a tax cut on GDP will be less than that for government spending. This is because a change in government spending affects GDP one-for-one, while part of a tax cut will be saved and will, at least initially, translate into a less than one-for-one increase in GDP.

Clearly, these results hinge on many underlying assumptions. One is that households are not assessing their future income when deciding how much to consume. Instead, they are assumed to spend a lot as long as their current income is high. However, households may be concerned about the impact of fiscal measures on their future tax bills. Households may not decide to consume as much if they expect taxes to rise and their future after-tax income to be lower. Moreover, this framework assumes that investment and net exports are insensitive to the change in fiscal policy. However, the response of investment will clearly depend on the behavior of interest rates, which in turn will depend on monetary policy. If monetary policy changes in response to fiscal policy, investment would be affected.

Large-scale econometric models often used in policymaking institutions make adjustments for household behavior and investment (see, for instance, Elmendorf and Reifschneider 2002). Nonetheless, the relative size of their fiscal multipliers is in line with this simple framework's predictions. For instance, earlier this year, Christina Romer, the chair of the Council of Economic Advisers, and Jared Bernstein, an advisor to Vice President Biden, estimated that the effects of permanently increasing government purchases by 1% of GDP would be to raise output by 1.5% two years after. At the same time, their model predicts that a tax cut of 1% of GDP would increase output by only 1% two years down the road.

Challenging the model

In a recent paper, Cogan et al. (2009) challenged the Romer/Bernstein estimates using an alternative New Keynesian model in which households and firms are more forward-looking than in typical large-scale econometric models. Using this model, the authors argue that a 1% increase in government spending would produce a mere 0.5% rise in output two years later.

In this framework, household and firm decisions to spend, invest, and produce are heavily influenced by their expectations of the future. Households anticipate that higher budget deficits will ultimately be financed with higher taxes, and they consume less as a result. Higher government spending thus crowds out consumption. Moreover, Cogan and his coauthors assume that, as the economy recovers following the increase in government spending, monetary policy becomes more restrictive, choking off investment. In contrast, Romer and Bernstein assume that the Federal Reserve keeps the federal funds rate constant, thus mitigating the adverse effect on investment. The crowding out of consumption and investment is relatively strong in the New Keynesian framework, offsetting much of the stimulatory impact of higher government spending.

In other words, the effects of fiscal policy on real GDP are quite sensitive to underlying modeling assumptions regarding the behavior of households, firms, and monetary policy. This creates fertile ground for good empirical work.

Recent empirical work

Empiricists interested in calculating the impact of movements in government spending and taxes on real GDP face multiple challenges, but the biggest hurdle is distinguishing fiscal policy changes that are fundamental from changes that are responses to economic conditions. Many influences other than tax and spending policy determine the trajectory of economic output. And taxation and spending vary over the course of the business cycle. The difficulty is to make sure to capture the effect of a change in fiscal policy on the economy and not the effect of changes in the economy on fiscal policy. Those fiscal policy changes that are independent of economic circumstances are called exogenous, and those that are reactions to economic conditions are called endogenous.

This is a particularly relevant issue because government spending and taxes respond endogenously to economic activity via automatic stabilizers--features built into the fiscal system to stimulate or depress economic activity automatically. Taxes automatically fall in recessions as household incomes decline. Transfer payments, such as unemployment insurance, rise. Moreover, government spending and taxes may have complex relationships with each other. For example, the payroll tax increased in 1965 to offset the costs of the new Medicare program on the federal budget (Romer and Romer 2008).

Typically, empirical studies adjust for the automatic stabilizers built into fiscal policy by taking into account movements in GDP when measuring government spending and taxes. Recent empirical analyses have taken a number of additional approaches to separate endogenous from exogenous factors.

Romer and Romer address the impact of tax changes by performing a narrative analysis of U.S. tax policy since 1945. Using the historical record, they try to isolate exogenous tax changes by identifying the key reasons underlying each modification to the tax code and rejecting those that were clear responses to economic activity. Alternatively, Blanchard and Perotti (2002) use a timing restriction to identify changes in government spending and taxes that are exogenous to unexpected movements in output. They argue that, because it takes time for legislators to understand a sudden movement in activity and then pass legislation to address it, it is reasonable to assume that, at high enough frequency, changes in taxes and government spending are independent of current output.

In contrast to these studies, Mountford and Uhlig (2005) use a mix of economic theory and time series analysis to identify exogenous movements in government spending and taxes. They build a small empirical model of the U.S. economy and look at the behavior of different "shocks" to that model, that is, disturbances that are unrelated to other variables in the system. They identify as exogenous movements in government spending those disturbances that end up raising government spending in the empirical model for a defined period of time. Similarly, exogenous movements in taxes are classified as those disturbances that end up raising tax revenues.

An interesting aspect of this new literature is that, notwithstanding their vastly different methodologies, they reach surprisingly similar conclusions. Regarding the impact of tax cuts on the level of real GDP one year after the change in taxes, the three studies predict a multiplier of roughly 1.2, as shown in Table 1. Moreover, Table 2 shows that, in contrast to theoretical predictions from the simple Keynesian framework, the analyses found that government spending had less bang for the buck than tax cuts. For instance, one year after the increase in spending, the impact on the level of real GDP is less than one-for-one, partly reflecting a decline in investment. There is more disagreement, however, about the effects of tax cuts on output two years after they are implemented, as Table 1 indicates. The analyses of Romer and Romer and Mountford and Uhlig find very large tax multipliers, while Blanchard and Perotti continue to find effects similar to those occurring after one year.

The stimulus package: Will it work?

Earlier this year, Congress passed a $787 billion fiscal stimulus package spread over 10 years. Of that total, $584 billion are spent in 2009 and 2010, with 19% of the funds allocated toward increases in government spending, 33.4% in transfers to the states, and 47.6% toward tax cuts. The findings from the three empirical studies, particularly those of Romer and Romer and Mountford and Uhlig, suggest that the fiscal stimulus package will boost growth substantially over the next two years, partly because it includes sizeable tax cuts that can be implemented quickly and that have significant effects on output.

Nevertheless, the uncertainty regarding those estimates remains high. Several economists remain skeptical that fiscal multipliers--whether from spending or taxes--are very large (see, for instance, Barro 2009). Moreover historical relationships may prove much less reliable during this downturn. Faced with a large decline in wealth and tight credit availability, households may very well respond differently to tax cuts today than they have in the past.

References

[URL accessed June 2009.]

Barro, Robert J. 2009. "Government Spending Is No Free Lunch." Wall Street Journal, January 22.

Blanchard, Olivier, and Roberto Perotti. 2002. "An Empirical Characterization of the Dynamic Effects of Changes in Government Spending and Taxes on Output." Quarterly Journal of Economics (November) pp. 1329-1368.

Cogan, John F., Tobias Cwik, John B.Taylor, and Volker Wieland. 2009. "New Keynesian versus Old Keynesian Government Spending Multipliers." Unpublished manuscript.

Elmendorf, Douglas W., and David Reifschneider. 2002. "Short-Run Effects of Fiscal Policy with Forward-Looking Financial Markets." National Tax Journal 55(3, September) pp. 359-386.

Mountford, Andrew, and Harald Uhlig. 2005. "What Are the Effects of Fiscal Policy Shocks?" SFB 649 Discussion Paper 2005-039.

Romer, Christina, and Jared Bernstein. 2009. "The Job Impact of the American Recovery and Reinvestment Plan." Manuscript.

Romer, Christina, and David Romer. 2008. "The Macroeconomic Effects of Tax Changes: Estimates Based on a New Measure of Fiscal Shocks." Manuscript.

---

Opinions expressed in this newsletter do not necessarily reflect the views of the management of the Federal Reserve Bank of San Francisco or of the Board of Governors of the Federal Reserve System.