Research on macro-prudential regulation: barely any effect
I promised last time a new post on macro-prudential regulation. Here it is. I recently reviewed six academic publications and noticed a curious fact: while most, if not all, of those papers conclude that some macro-prudential policies can have some effects on some asset prices or credit growth, they remain quite careful in their conclusions and are often aware of the limitations of their methodology. However, when other papers refer to them (usually in their literature review), they always overemphasise the apparent successes of those policies and avoid mentioning their shortcomings and limitations almost entirely. Consequently, a reader that would only open one of those research papers would only see macropru under an overwhelmingly positive light.
But, don’t be mistaken, it is not what most of those researchers said. Their conclusions are usually much more nuanced than the depiction and oversimplification we hear in the media and in the mouth of other academics and central bankers.
The first paper, published in 2011 by IMF staff, titled Macroprudential Policy: What Instruments and How to Use Them? Lessons from Country Experiences, was an early attempt at looking at the economic effects of macropru. While thick, the paper is clearly not comprehensive and a number of its findings were in opposition with findings of later research.
Running a cross-country regression analysis on data from 49 countries, they found that a number of macropru instruments were effective in dampening procyclicality. Most ‘effective’ instruments were targeting particular products or asset classes (i.e. LTV or DTI caps), although they did find that fluctuating reserve requirements helped, in contradiction with other papers.
Importantly, they stress that there are ‘costs’ involved in using those instruments. That is, potentially lower economic growth or ‘unintended distortions’. They also point out that macropru could be subject to regulatory and cross-border arbitrage and show that discretionary decision-making, rather than rules, dictate the use of those policies. Finally, they also warn against using macropru as an alternative to monetary policy in order to address ‘excess demand’, and subsequent research has indeed shown that macropru is no substitute for monetary instruments. They do not research any of those issues.
Debunking the myth that macropru was a ‘new’ tool for central bankers, they show that many, many countries has already utilised them before the crisis.
My main issue with this paper is that, similarly to another macropru research paper I reviewed in a previous post, it takes correlations way to literally. For instance, it says that macropru was successfully used in China to limit credit and house price growth, and in Spain to help cover the large losses of the financial crisis. Well, it seems to me that, those ‘successes’ have been rather poor given how severely Spanish banks suffered during the crisis and how rapidly credit growth has been since the implementation of those measures in China… I would at best call this a ‘quarter-success’, and this, even without even looking at the potential credit allocation distortions macropru has brought about in those countries.
Moreover, their correlations are not fully convincing: from some of their charts, it seems that the implementation of a macropru measure was sometimes followed by no effect for a while, or in other cases, that the implementation only occurred after a change in credit growth trend (see chart below).
The second paper, published in 2012 also by IMF staff, is titled Macroprudential Policies and Housing Prices – A New Database and Empirical Evidence for Central, Eastern, and Southeastern Europe. This one is more targetted, both in term of geography and in terms of target (housing prices). Here again they show that many countries had been using macropru for quite some time, and that it didn’t help them much preventing crises.
They find that some macropru measures, such as raising minimum regulatory capital ratios, did have an impact on house price growth, while others, such as increasing reserve requirements or dynamic provisioning, did not. However, those effects only slow down the housing cycle rather than stop or reverse it, and more importantly, they are temporary. Growth seems to reaccelerate a few months after their introduction. Their study does not look beyond this short timeframe.
Curiously, and in contradiction with a later study (see below), they find that easing regulatory capital requirements helped dampen the effects of a crisis.
The authors also show that different countries demonstrate different macropru ‘activism’ (see chart below), and that, as expected, a number of measures adopted were procyclical, amplifying the crisis. Clearly, this is a major weakness for macro-prudential policies.
Overall, this paper does not look at whether macropru leaks in a way or another, but does show that its effectiveness is limited and temporary, and that it is clearly subject to usual problems of discretionary policies and central planning.
The third paper, published in 2013 by BIS staff, is titled Can non-interest rate policies stabilise housing markets? Evidence from a panel of 57 economies. Like the previous one, the scope of this paper is limited to housing markets. But at least it allows us to easily compare their results in order to check the consistency of those findings.
The authors used two types of regressions to assess the effects of certain types of macropru tools affecting the supply side (reserve/liquidity requirements, limits on credit growth, risk weighting and provisioning requirements, limits on housing sector exposure) or the demand side (cap on LTV and debt service to income and housing-related taxes).
What did they find? Not much:
From a policy perspective, the negative results are in some respects as important as the positive ones. One such finding is that instruments affecting the supply of credit generally by increasing the cost of providing housing loans (reserve and liquidity requirements and limits on credit growth) have little or no detectable effect on the housing market. Nor do risk weighting and provisioning requirements, which target the supply of housing credit. Exposure limits, which work not by the cost of lending but through the quantity restrictions on banks’ loan supply, may be an exception in this regard, although the small number of documented policy actions makes it hard to draw firm conclusions. Measures aimed at controlling credit supply are therefore likely to be ineffective.
Of the two policies targeted at the demand side of the market, the evidence indicates that reductions in the maximum LTV ratio do less to slow credit growth than lowering the maximum DSTI ratio does. This may be because during housing booms, rising prices increase the amount that can be borrowed, partially or wholly offsetting any tightening of the LTV ratio. None of the policies designed to affect either the supply of or the demand for credit has a discernible impact on house prices. This has implications for the degree to which credit-constrained households are the marginal purchasers of housing or for the importance of housing supply, which is not explicitly considered in this study. Only tax changes affecting the cost of buying a house, which bear directly on the user cost, have any measurable effect on prices.
In other words, apart from increasing taxes on housing (which isn’t really macropru), none of the macro-prudential instruments studied had any actual effect on housing prices. This doesn’t sound very positive. Moreover, due to the narrow scope of the paper, it doesn’t look at potential distortions resulting from the same policies.
Interestingly, the paper provides a quite comprehensive database of macro-pru and tax-related measures taken since the 1980s (see below). After seeing this, I’m really unsure how macropru proponents can still claim this type of regulation is new and effective.
The fourth paper (another IMF staff one) was published in 2014 and is titled Macro-Prudential Policies to Mitigate Financial System Vulnerabilities. Looking at a database of 2,800 banks in 48 countries, they analyse the balance sheet response to various types of macro-prudential tools during the 2000-2010 period. A quick remark first of all: they seem to only use USD balance sheet figures, which I find surprising given that year-on-year FX fluctuations would affect their conclusions. They don’t seem to be controlling for this effect, but I might have misread.
in particular caps on borrower and financial institutions assets and liabilities–based measures to be effective, while buffer-based policies seems to have little impact on asset growth. Overall, there is little evidence that the effectiveness of these tools varies by the intensity of the cycle.
Caps on borrowers (LTV, DTI…) are confirmed to have an effect, in line with some of the studies above (although the end goal remains to control housing price growth, and the previous study seems to show that this was not happening). However, the effect is small to say the least: banks’ asset growth was reduced by 0.44% in average. More surprisingly, they find that countercyclical capital buffers not only have little effects on credit growth, but also
do not help to provide cushions that alleviate crunches during downswings. As such, macro-prudential tools may be less promising to mitigate adverse events.
This is a major issue for regulators as countercyclical buffers and dynamic provisioning have been mostly advertised as helping alleviate a downturn by making banks more solid, thereby providing continuity to the flow of credit to the economy.
Finally, they find that macropru has been mostly used by closed capital countries, which have less liberalised financial systems. This makes sense as there are then fewer opportunities for those measures to ‘leak’. But this has not prevented many of those developping countries to experience crises anyway.
More recently (end-2015), BoE staff published a piece titled Regulatory arbitrage in action: evidence from banking flows and macroprudential policy. The piece is interesting because it demonstrates how regulatory arbitrage within a macro-prudential framework. It is based on a database of banking flows in 37 countries (but not on individual banks data), between 2005 and 2014.
They find that macropru measures that do not apply to all banks within a certain country suffer from leakages. This is possible because, apart from countercyclical buffers, Basel 3 do not impose cross-border regulatory coordination. Consequently, macropru tools that affect capital levels only apply to domestic banks and domestic subsidiaries of foreign banks, but not to branches of foreign ones. The result is an increase in foreign borrowing by domestic corporates when capital constraints are tightened for the domestic banking system.
However, when macropru tools (such as tigthening of lending standards) apply indiscrimenately to all banks within a given jurisdiction, they find no evidence of this cross-border borrowing (the charts they present aren’t as clear cut, even though the effect is clearly weaker – see below). Evidence is mixed when reserve requirements are varied.
More worryingly, their conclusion points to macropru regulation as a major driver of cross-border capital flows through regulatory arbitrage:
This suggests that uneven application of regulation may be a driver of international capital fows. This is over and above the effect, which has been documented extensively in the literature, where banks transfer funds to markets with fewer banking regulations. Our results show a subtle contrast: banks transfer funds to countries which tighten regulatory standards, but transfer these funds when the regulatory tightening does not apply to them, and instead confers upon them a competitive advantage.
Finally, the latest paper I rewiewed, published in January 2016 by BIS staff, is titled Is macroprudential policy instrument blunt?. The paper seems to suffer from a ‘correlation does not imply causation’ issue. It focuses on Japan from the 1970s to the 1990s, as local authorities used a policy instrument called ‘quantitative restriction’ (QR), which requires banks to reduce their real estate lending.
I won’t spend long on this paper as I believe it doesn’t bring much to the literrature given its reliance on what seems to be spurious correlations. We all know the booms that Japan experienced in the 1980s. We also all know that when this boom turned into a burst, the whole economy collapsed. Does this have anything to do with the implementation of QR? Unlikely. Yet this is what the researchers who wrote this paper seem to say, as they write that “QR affected the aggregate economy by damaging the balance sheet of banks and non-financial sectors”. Unless it was the crisis.
In the end, it does look like the evidence that macro-prudential policies are effective is rather inconclusive. Some seem to have minor effects, temporarily, providing one does exclude potential leakages and distortions and believes in the timely implementation of those tools by regulators. Others seem to have no effect whatsoever. And worse, academics seem to reach different conclusions regarding the exact same instruments.
Yet, the literrature review section of many of those papers depicts macropru in rather positively. Take the fourth paper listed above. Referring to the first paper listed in this post, it says that its authors “document evidence of some policie being effective in reducing the procyclicality of credit and leverage.” Well, I had a slightly different reading.
To be fair, they do also report the relatively negative results of some other studies (such as the third one above). This is in stark contrast with most central bankers and regulators, who describe macro-prudential regulation under an overwhelmingly positive light, and who would like us to believe that enlightened policymakers manipulating macropru can solve all the problems that their own misguided monetary policy creates, independently of any Public Choice issue. Let’s not believe.