VoxEU Column Monetary Policy

Macroprudential tools, capital controls, and the trilemma: Insights from the Bretton Woods era

In the Bretton woods system, capital controls ensured the independence of monetary policy. This column argues that it is impossible to understand how they worked without understanding their role in supporting credit controls at the time, which were used to fight inflation without raising the domestic interest rate. This may be relevant today in emerging markets in which central bank instruments still resemble those used under Bretton Woods. 

Assessing how domestic monetary policy is constrained by international capital flows remains a key but open empirical and theoretical question (Farhi and Werning 2014, Shambaugh and Klein 2015, Rey 2016, Obtsfeld et al. 2017). It is often approached from the angle of the trilemma – the impossible trinity of international finance – which stipulates that only capital controls are capable of ensuring the autonomy of monetary authorities' actions in a system of fixed exchange rates. The Bretton Woods system (1946-1971) often occupies a special place in these debates, because it remains the paradigmatic case of an international monetary system where exchange rates were fixed, but where the independence of monetary policy was ensured by generalised capital controls (Ghosh and Qureshi 2016, Obstfeld and Taylor 2004, 2017). It is not surprising that the Mundell-Fleming model was framed during this period. The Bretton Woods system is also a reference point in the discourse of policymakers, used as evidence that institutionalised central bank cooperation and capital controls can provide solutions to the constraints of international finance.1

However, our historical knowledge of monetary policy under Bretton Woods remains very limited. Moreover, we still know little about the functioning and logic of capital controls during this period and their interaction with central bank policies. In two papers, I argue that it is impossible to understand the role of capital controls and to measure the autonomy of monetary policy under Bretton Woods without understanding the role played by credit controls at the time (Monnet 2017, 2018). Quantitative credit controls (i.e. either ceilings on banking credit growth or rationing at the central bank discount window) were the main tools of the domestic countercyclical policies of the major central banks in Western Europe and Japan until the 1980s. They were used to fight inflation without raising the domestic interest rate. The domestic interest rate could remain in line with the international rate. In the words of Milton Gilbert, the BIS's renowned economic adviser in the 1960s, “the objective was to assign the interest rate to the external side and to find ways of managing the rate of domestic credit expansion without relying wholly on interest rates.”2

In this context, capital controls were not actually necessary to maintain a spread between domestic and international rates, but they were a key instrument for making credit controls effective. Capital controls prevented banks and private agents from circumventing domestic credit controls.

This historical experience should form the basis for the current discussion on the link between macroprudential policies and capital controls (Ostry et al. 2011, Korinek and Sandri 2016), since the credit controls used by central banks during the Bretton Woods period have some technical similarities with current macroprudential tools designed to limit the growth of credit. 

From a more technical perspective, the Bretton Woods experience usefully reminds us that the difference between national and international interest rates is not a good measure of policy autonomy when central banks use quantitative credit controls to stabilise inflation, exchange rates, or credit growth. 

Disconnect between quantitative credit controls and interest rates

An important assumption underlying standard models in international macroeconomics is that interest rates are the main instruments of monetary policy, or at least reflect the stance of monetary policy, so that they increase when money supply decreases, as in the Mundell-Fleming model. The purpose of capital controls is to allow monetary authorities to maintain a spread between domestic and international interest rates.  However, the central bankers of the 1950s and 1960s had another mechanism in mind. The alternative solution required central banks to disconnect quantities and prices: central banks used a variety of credit controls and reserve requirements to fight inflation or, on the contrary, increase output, while keeping their leading interest rate stable (usually the discount rate or the Lombard rate). Thus, the domestic interest rate remained consistent with international rates, and central banks enjoyed the necessary autonomy to achieve their national inflation and unemployment targets. A 1960 speech by the Governor of the Bank of Italy, Donato Menichella, highlighted precisely this mechanism: 

“The measures [credit rationing] we adopted were of a quantitative nature, so that we were able to avoid a change in the discount rate. […] In avoiding a rise in Bank rate, our tactics were in line with the spirit of the policy which, as I said earlier, tends to dissociate measures aiming at domestic monetary equilibrium from those which, with sometimes contrary effects, influence international capital movements.”3

It is well known that credit controls (limits on banking credit growth) and reserve requirements (cash deposits with central banks or wider liquidity requirements) were the main instruments of most central banks during this period, but the consequences of using these tools for escaping the trilemma, and the interaction between credit controls and capital controls, have been underestimated in recent research.  

In the early 1960s, several European countries such as West Germany, Italy, and France relied – to various extents – on the disconnection between interest rates and quantitative credit controls to combat inflationary pressures while avoiding worsening their balance of payments surpluses (Monnet 2017, 2018). Many contemporary economists noted that these mechanisms provided additional justification for the use of quantitative credit controls and reserve requirements by central banks (see references listed in Monnet 2017). The willingness of monetary authorities to intervene in the allocation of credit, consistent with other activist state policies in the economy, such as export subsidies and industrial policies, also justified credit control.

Financial regulation, capital controls, and interest rate spreads

Credit controls were accompanied by strict banking regulations. In fact, credit controls took the form of banking regulations used for monetary policy purposes. The reason why a decline in credit and money in the economy was not transmitted to market interest rates is that there were few market interest rates during this period. Most loan and deposit rates were regulated and financial systems were highly segmented, with some institutions specialising in loans to specific sectors at subsidised regulated rates. When interest rates on government bonds rose due to credit restrictions, central banks were eager to buy these bonds to keep rates low. 

Certainly, capital control played a role in such a framework. But part of the role was different from that explained by the Mundell-Fleming model.  In the standard model, capital controls allow countries to maintain a gap between national and international rates, so that a high spread is interpreted as evidence of binding capital controls (Obstfeld and Taylor 2004). In practice, the alternative policy we described was also based on capital controls, but for another reason. Capital controls were complementary to credit controls. Without capital controls, restrictive limits on bank credit growth would not have been effective in stabilising inflation, as firms could simply have borrowed funds abroad to circumvent controls. It is important to bear in mind this other raison d'être of capital controls, as it has different implications for interest rate differentials: when capital controls were used to make credit controls effective, the spread between domestic and international interest rates remained close to zero.  This is what we observed, for example, in France in 1963-65, when credit controls and liquidity requirements were imposed to combat inflation without increasing the discount rate, while capital controls on capital inflows were tightened (Figure 1).  The spread between French and US interest rates did not increase during this episode, while the French central bank implemented an anti-inflationary monetary policy with a combination of credit and capital controls but without changing interest rates.

Figure 1 Restrictive monetary policy, credit controls and the spread between US and French rates, 1958-1968

Sources: FRED for the Fed fund rate; reports of the CNC for the Frence rate; IFS for spot and forward exchange rates.
Notes: The deviation from the covered interest rate parity is d = F/S(1 + i) – (1 + u) where F and S are the forward and spot exchange rates, I and u are the French and US money market rates. Data are quarterly such that we use the average value of call money rates over three months. The forward exchange rate is the one at the end of the previous month. Grey shaded areas are restrictive episodes of monetary policy when ceilings on credit growth were in place.

Conclusions for today’s monetary and macroprudential policies

The justification for capital controls under the Bretton Woods system went well beyond the standard justification provided by the Mundell-Flemming model.  Capital controls were an essential element of directed credit policies, typical of what has been referred to as ‘embedded liberalism’ (Ruggie 1982) or, from a different perspective, ‘financial repression’ (Reinhart and Sbrancia 2015). Capital controls were also an essential means of making monetary policy effective at a time when central banks relied much more on quantitative credit controls and reserve requirements than on interest rates and open market operations. 

What conclusions should we draw? The most obvious one today concerns emerging markets where central bank instruments still resemble those used in Western Europe and Japan under Bretton Woods. At least historical experience should take us beyond the mechanisms described in the Mundell-Fleming model and take into account how the combination of capital and credit controls offers flexibility to central banks, particularly those facing both large capital inflows and rising inflation. 

A second conclusion concerns what is now called macroprudential policy. Since current macroprudential rules aim to smooth credit cycles, it is tempting to interpret past experience in credit supervision as a revealing precedent (Kelber and Monnet 2014). It should be borne in mind, however, that in the Bretton Woods environment, characterised by pervasive financial regulation and in which banks provided the bulk of credit to the economy, central banks used quantitative limits on credit growth to combat inflation rather than to achieve financial stability objectives.  That said, the potential disconnect between interest rates and the stance of quantitative credit controls still exists. Therefore, the mechanisms described above are potentially still relevant today, particularly in financial systems that are highly regulated. This historical experience shows how important it is to examine the interactions between capital controls and macroprudential tools (Korinek and Sandri 2016) and between the monetary policy trilemma and the financial trilemma (Obstfeld and Taylor 2017).

Authors’ note: The opinions and arguments expressed here are those of the authors and do not reflect the views of the Banque de France or the Eurosystem.


Farhi, E and I Werning (2014), “Dilemma not trilemma? Capital controls and exchange rates with volatile capital flows”, IMF Economic Review 62(4): 569-605. 

Ghosh, A R and M Qureshi (2016), “What’ s In a Name? That Which We Call Capital Controls”, IMF Working paper. No. 16/25. 

Kelber, A and E Monnet (2014), “Macroprudential policy and quantitative instruments: a European historical perspective”, Financial Stability Review 18: 151-160.

Klein, M and J Shambaugh (2015), “Rounding the Corners of the Policy Trilemma: Sources of Monetary Policy Autonomy,” American Economic Journal: Macroeconomics 7(4): 33-66.

Korinek, A and D Sandri (2016), “Capital controls or macroprudential regulation?”, Journal of International Economics 99: S27-S42.

Monnet, E (2017), “Credit controls as an escape from the trilemma. The Bretton Woods experience”, CEPR Discussion Paper 12535, forthcoming in the European  Review of Economic History.

Monnet, E (2018), Controlling Credit. Central Banking and the Planned Economy in Postwar France,1948–1973, Cambridge University Press, forthcoming.  

Ostry, J, A Ghosh, M Chamon and M Qureshi (2011), “Capital Controls: When and Why?” IMF Economic Review 59(3): 562-580.

Obstfeld, M, J D Ostry and M S Qureshi (2017), “A tie that binds: Revisiting the trilemma in emerging market economies”, Review of Economics and Statistics. 

Obstfeld, M and A M Taylor (2004), Global capital markets: integration, crisis, and growth, Cambridge University Press.

Obstfeld, M and A M Taylor (2017), "International monetary relations: taking finance seriously." Journal of Economic Perspectives 31(3): 3-28.

Reinhart, C M and B Sbrancia (2015), "The liquidation of government debt," Economic Policy 30(02): 291-333

Rey, H (2016), “International Channels of Transmission of Monetary Policy and the Mundellian Trilemma,” IMF Economic Review 64(1): 6-35

Ruggie, J G (1982), "International regimes, transactions, and change: embedded liberalism in the postwar economic order", International Organization 36(02): 379-415.


[1] Such different people as Zhou Xiaochuan, Paul Volcker or Yanis Varoufakis, for example,  have recently called for a ‘new Bretton Woods’.

[2] See Monnet (2017) for references.

[3] See Monnet (2017) for references.

2,835 Reads