America has made frequent use over the last 100 years of financial regulatory policies designed to control credit cycles—what we now call cyclical “macroprudential” policies. This contradicts a common belief in policy circles, based on their lack of use in the last two decades, that these tools were never used in the U.S. I recently co-authored a comprehensive paper on American macroprudential policy since 1913 (Elliott, Feldberg, Lehnert (2013), or EFL (2013) for short). The paper provided a taxonomy and catalog of these historical actions, an explanation of the circumstances and politics affecting the significant actions, and some initial statistical analyses of the impacts of the policies. However, it did not suggest policy implications for the future, except indirectly.
This paper builds on EFL (2013) by drawing a number of lessons for American macroprudential policy going forward. These conclusions are mine alone and may or may not agree with the views of my co-authors of the earlier paper, although I am indebted to them for many insights and for their comments on a draft of this paper.
The key lessons that I draw are:
- Macroprudential policies are feasible in a U.S. context
- Political support can be obtained for macroprudential tightening
- Macroprudential and monetary policy blend together
- Cyclical macroprudential policy can affect credit supply as intended
- Different economic sectors can be targeted
- Macroprudential policy may be easiest when measures appear technical
- A major mistake can make future macroprudential policy much harder
- Macroprudential policy may be most easily done through a single body
Overall, U.S. history strongly suggests that it is feasible and desirable to utilize macroprudential policy more actively in the future in America.
Former Brookings Expert
Macroprudential policies are feasible in a U.S. context
The extensive catalog of historical actions in EFL (2013) demonstrates that it is possible in an American political and economic context to use macroprudential policies to try to reduce the damage from booms and busts in credit markets. Since 1913, we counted 245 instances of the use of tools that we considered to be of a macroprudential nature and intended to counter credit cycles. (Macroprudential policy can also be of a structural nature, intend to strengthen the system across time periods. We have not focused on these uses.)
This long time period encompassed a wide range of political and economic conditions, including control by each of the major political parties. Thus, it seems unlikely that there is some uniquely American feature that would prohibit the use of macroprudential policies in the future. This is an important point, because there has been an assumption by many people that the U.S. had seldom used macroprudential policies because they simply did not fit the American ethos and circumstances.
Political support can be obtained for macroprudential tightening
A concern about cyclical macroprudential policy is that political constraints may render it impossible to tighten policy. (As with monetary policy, it is presumed to be easier to persuade people to accept looser credit conditions.) For example, many doubt that the authorities would have been allowed to implement a tightening in credit conditions during the boom leading to the recent financial crisis. This is a very reasonable concern, both here and in other countries, and there will doubtless prove to be constraints of a political nature. Indeed, in designing the Basel III capital framework, global policymakers implicitly recognized this point when they favored a rule-based trigger for implementation of the countercyclical capital buffer. Under the framework, if national credit-to-GDP growth sufficiently exceeds its long-term trend, authorities must either turn on the countercyclical buffer or explain why doing so is inappropriate.
Nonetheless, the U.S. has implemented substantial tightening measures in response to some previous credit booms. There has always been political opposition, but there has often been sufficient political support to uphold the measures. The most striking instance was the use of powerful credit restraints in 1980 as part of a campaign against inflation, described below. Another important example took place in the early 1950’s, with a coordinated response by multiple regulators to counter a housing boom that appeared excessive. It is not that there was an absence of political pushback, but rather that the rationale for action appeared compelling enough to allow substantial actions to proceed.
It is worth noting that some of the most potentially effective tools, such as Loan to Value limits for mortgages, may be the hardest to push through because of political opposition. The chances of being able to do so are heightened if the tools can be readied in advance of when they are needed, so that the process can be quicker and there are fewer points at which politics can intervene. EFL (2013) shows that supervisors in the U.S. had great difficulty even putting out formal guidance that might have slowed down the excessive boom in commercial real estate. It would likely have been easier to act if better tools had already been created in anticipation of such a need.
Macroprudential and monetary policy blend together
We were faced with an interesting fundamental task in researching EFL (2013), which was to determine when a policy was of a cyclical macroprudential nature during the many decades prior to the invention of the term “macroprudential.” This required some judgment calls, but I believe that the choices we made are generally reasonable, consistent, and useful. However, it was striking how often the question arose as to whether an action was in the furtherance of monetary policy or was macroprudential in nature.
Monetary policy works very largely via interactions with and influences upon financial institutions and markets. So does macroprudential policy. Monetary policy attempts to affect credit conditions, as does macroprudential policy. The rationale and methods generally differ, but there is definitely overlap. The most difficult call was on the use of reserve requirements. For many decades the Federal Reserve (Fed) has required member banks to keep reserves as deposits at the Fed proportionate to the size of their total deposits and often that of other specific liability categories. The percentages that have needed to be held have been increased and decreased over time in order to tighten or loosen credit conditions.
As we indicate in EFL (2013):
“Most treatments of reserve requirements describe them as tools of monetary policy, along with open market operations and discount rates. Indeed, in the U.S., by setting a floor on the volume of reserves banks have to hold, reserve requirements influence the demand for federal funds and thus the equilibrium federal funds rate. Nonetheless, we include them in our taxonomy of macroprudential tools because they can, in principle, exert a direct effect on the supply of loans, while a change in the target federal funds rate affects both the supply and demand for loans. In practice, policymakers, even in modern times, have described changes in reserve requirements as motivated by a desire to control credit supply independently from the setting of monetary policy. For example, the 1992 decrease in reserve requirements was part of a coordinated program of government actions to ease what were perceived at the time to be excessively tight credit conditions.”
There are a number of instances described in EFL (2013) in which reserves requirements were used either for the purpose of altering overall credit conditions or, sometimes, to alter credit conditions more selectively.
Muddying the waters further, the Fed’s ability to determine interest rates was severely constrained during most of World War II and through 1951. Until the so-called Treasury Accord of that year, the Fed was obliged to ensure that Treasury securities could be sold at low interest rates, in order to assist the war effort and to help deal with the huge overhang of debt that existed for years after the war.
During this period, the Fed relied more heavily on the other tools available to it, including macroprudential instruments such as credit controls, but especially reserve requirements, which fall in the borderland between the two types of policies.
There are some analysts and policymakers who wish to use macroprudential policies to ensure financial stability and hope that monetary policy would not need to be set with financial stability in mind. The very muddy distinction historically between these types of policies suggests that life will not be quite this simple.
In framing the distinction between monetary and macroprudential policy, it is useful to consider the differing aims of the two. Monetary policy in the U.S. is required to meet the dual mandate of low inflation and full employment and therefore focuses on the classic business cycle. Macroprudential policy can be thought of as striving to maintain financial stability by moderating financial cycles. The official business cycle dating committee of the National Bureau of Economic Research counts 11 business cycle peaks from 1945 to the present. Although no similar authority exists to define financial crises, over the same period, by any definition; there have clearly been far fewer financial crises. Research at the Bank of England has reached the same conclusion that business cycles are substantially shorter on average than financial cycles.
Cyclical macroprudential policy can affect credit supply as intended
The statistical analysis presented in EFL (2013) shows that cyclical macroprudential policy actions did indeed affect credit supply in significant ways. Because the tools used and credit aggregates targeted changed over the years covered by our study, our statistical evidence is effectively a series of case studies. Indeed, by the nature of macroprudential policymaking, the fundamental structure of the financial system evolved in response to policymaker actions, suggesting caution in using the precise quantitative estimates from our study in forecasting the effect of similar policies today. Despite this caveat, some themes appear to be fairly convincing from the data.
Macroprudential policies affected credit growth. The statistical tests that used all macroprudential policies taken together did not reach statistical significance at a high level of significance, reflecting the shifting toolkit and targets mentioned above, as well as some measurement difficulties. However, tightening actions were associated with a decrease in credit and easings with an increase. The more precise tests that looked at changes in reserve requirements showed that a tightening generally led to a 1% decrease in bank credit.
The policies were less clearly effective in “pushing on a string”. As with monetary policy, it seems to be harder to spur increased lending in pessimistic times than it is to restrain lending in optimistic times. The statistical analysis of the effect of reserve requirements did not produce a clear relationship between easings and increases in bank credit, unlike the results for tightenings.
Some lending moves away from banks or other restrained entities. As stated in EFL (2013), the statistical analysis of the effects of reserve requirements shows that “[t]otal consumer credit, which includes loans made by non-banks, falls less. This suggests either that non-bank lenders stepped in to make loans to households following a tightening in reserve requirements, or business lending fell by more than consumer lending in response to a tightening. Because reserve requirements operate directly on banks, it is at least plausible that some lending leaked outside the banking system in response to higher reserve requirements, although we cannot verify this directly.”
We did not measure the effect on systemic resilience, which could be more crucial. It is important to emphasize that EFL (2013) did not attempt to capture what could be the most critical advantage of cyclical macroprudential policy in the future. Whatever degree of benefit exists from restraining credit cycles by influencing the volume of lending, it may be more crucial to ensure that financial institutions are better prepared to deal with the consequences of a credit bust following a boom. If the quality of mortgages has been increased by tightening credit standards, or the ratio of bank capital to assets has been raised during the expansion, for example, then the harm done when a boom collapses may be considerably reduced.
Different economic sectors can be targeted
The ability to execute macroprudential actions in the U.S. does not appear to be limited to a particular economic sector, nor to the economy as a whole. The authorities have acted in ways designed to affect borrowing across the economy as a whole, as well as sometimes targeting specific sectors. These sectors frequently included housing or other forms of consumer borrowing. Business borrowing was less frequently targeted, but such targeting did occur.
Macroprudential policy may be easiest when measures appear technical
It is noteworthy that the great bulk of macroprudential actions were taken through measures such as the reserve ratio or margin requirements that the public does not focus on or particularly understand. The actions may be powerful, but they are indirect and appear to be of a technical nature. Both factors likely make it harder for political opposition to develop.
That said, there have been times when actions were taken that would receive more attention from the public. The most notable was the invocation of credit controls in 1980 which were launched in a major campaign and which affected a wide variety of activities. More frequently, down payment or other requirements on mortgage loans were altered. The public understands down payment requirements and much of the public cares about the topic.
The balance of historical activities suggests that some of the actions that may be easier to take in the future would include:
- Countercyclical capital
- Countercyclical risk weightings
- Countercyclical loan loss provisioning
- Countercyclical liquidity buffers
- Countercyclical margin and collateral requirements
A major mistake can make future macroprudential policy much harder
The ability of macroprudential authorities to act can be curtailed sharply and quickly in the event of a major mistake in this policy area. This is illustrated very clearly by the history of the Credit Control Act. Congress in 1969 gave the President the power to direct the Fed to implement credit controls in the U.S. economy, with a very wide grant of authority. This was not used until 1980, when President Carter invoked these powers to induce the Fed to take strong actions to rein in credit growth, which was seen as contributing to the inflationary environment. (See EFL (2013) for a considerably more detailed explanation of the actions and the surrounding circumstances.)
The economy quickly sank into recession and there appeared to be a very direct connection between the credit controls and this drop in activity. Once the controls were removed, which was quite quickly thereafter, economic growth resumed, with a considerable bounce-back. The disastrous use of such a strong set of macroprudential tools made it much harder to attempt future macroprudential actions, even of milder and more conventional form. It did not help that President Carter’s subsequent defeat in the 1980 elections may have resulted in substantial part from these unfortunate economic conditions. Within two years, Congress repealed the authority to establish credit controls. In general, 1980 may be seen as the high point in the use of macroprudential policies, with a very substantial drop starting soon thereafter and a virtually total cessation over the next decade.
Clearly there were other changes in the political, economic, and social environment after 1980, particularly the tide of deregulation that began in the 1970s and continued with the “Reagan Revolution”, but it seems highly likely that the disaster of 1980 played an important role in the decline of macroprudential policy. Certainly opponents of such policies cited it frequently.
This lesson has particular relevance for policymakers in the U.S. today, because America is essentially making a fresh start on macroprudential policy after the recent financial crisis and the passage of the Dodd-Frank Act. Little was done in this area for decades, so those actions that are taken will receive particular scrutiny. Global views on how macroprudential policy fits into the larger scheme of things have also evolved very considerably, adding to the sense of a fresh start. Should the U.S. authorities make a large mistake in this area, or be perceived to have done so, it could halt significant macroprudential policy of a cyclical nature in the U.S. for many years. This may be counteracted over time by the successful use of such tools in other countries, but home grown lessons tend to have much greater force, especially political force, than lessons from abroad.
This point is especially important because macroprudential policy is not solidly based in political and public support at this point. There is a very strong consensus for monetary policy to be undertaken, although there does remain a core of support for a gold standard. Monetary policy mistakes are unlikely to lead to the abolition of the Fed. The equivalent cannot be said for macroprudential policy.
Macroprudential policy may be most easily done through a single body
A large majority of the macroprudential actions were taken by a single entity, the Federal Reserve, although there are a number of instances in which other entities acted, often in combination with the Fed. For instance, fears of excessive growth in housing credit in the 1950’s led to aligned moves by the Fed, the Federal Home Loan Bank Board, and the Federal Housing Administration.
It is clearly easier for a unified body to take a step than it is to coordinate among a number of players. Largely for this reason, the IMF has argued on the basis of a global analysis that “the central bank needs to play an important role” in macroprudential policy, and has found that policy responses that involve the central bank tend to be quicker. However, U.S. history shows that it is not impossible to reach a consensus, even for tightening moves, which are inherently more difficult to sell to politicians and the public.
Nonetheless, I have serious concerns that our macroprudential coordinator, the Financial Stability Oversight Council, will find it quite difficult to act, because of its nature as a large committee representing many regulatory interests. U.S. history does not present definitive evidence in this regard, but the Fed’s predominant role does suggest that a more unified entity might have better luck implementing macroprudential policy.
U.S. history strongly suggests that we can and should use macroprudential policy more actively in the future, after the hiatus of several decades prior to the financial crisis. We may do so most effectively by heeding the lessons of the last century of experimentation in this country.