Recent central bank interventions following the Global Crisis have raised new interest in quantitative measures as instruments of monetary or macroprudential policy (Borio 2011, Galati and Moessner 2013). In fact, quantitative controls – especially credit controls – have been used as primary tools of monetary policy for decades in western Europe and east Asia, usually during periods when these countries were experiencing their highest ever rates of growth. Many countries, including Brazil, India, and China, still use them today.
Despite their importance, these tools remain largely absent from the standard literature on the effects of monetary policy. Since traditional econometric methods usually consider interest rates to be the primary instrument of monetary policy, it is difficult to compare the effectiveness of quantitative controls with the standard results obtained by Sims (1992) and Christiano et al. (1999) concerning conventional monetary policy.
French post-war monetary policy from 1948 to 1973 was a paradigmatic example of the use of temporary quantitative credit controls that nearly eliminated the role of interest rates. In Monnet (2014), I provide a detailed analysis of the way credit controls and other quantitative tools were used in France, and I propose a new method to assess their macroeconomic effects.
The first contribution of the paper is to demonstrate that an effective way to assess the stance of monetary policy when interest rates are not the primary instrument is to follow a narrative approach (Friedman and Schwartz 1963, Romer and Romer 1989), that is, to examine archival evidence directly concerning policymakers’ intentions and decisions. No reliable quantitative indicator exists concerning French monetary policy from 1948 to 1973, since the Central Bank had to change its instruments constantly to adapt both to financial innovation and to the circumvention of previous sets of instruments by the banks.
The second contribution of this paper is to combine the narrative approach with a vector autoregressive (VAR) estimation to demonstrate that quantitative controls on credit and money had a strong influence on nominal and real variables, but not on interest rates. There was a complete disconnect between quantities (of money and credit) and prices (interest rates).
In what follows, for simplicity, I use the term monetary policy to refer to the whole set of central bank operations but these operations also aimed to affect credit allocation (Monnet 2013).
Quantitative instruments and the problem of measuring the monetary policy stance
From 1948 to 1973, the Bank of France used more than 15 tools to control directly banking credit or liquidity. None of them has been used over the whole period (see Table 1). The main ones were bank-by-bank rediscount ceilings (a cap on the amount of bills that banks can refinance/rediscount at the Central Bank), credit ceilings (a quantitative limit to the growth of banking loans), liquidity ratios (banks were forced to hold a fixed proportion of non-liquid assets, such as long-term bonds) and reserves requirements (obligatory reserves at the Central Bank, as a proportion of deposits or of outstanding loans). Rediscount ceilings and liquidity or reserve ratios were used on a continuous basis, with the Bank changing their values when it wanted to expand or restrict money and credit creation. By contrast, credit ceilings – in use from 1958 to 1973 – were imposed only when the Bank decided to make the monetary policy stance more restrictive, and were lifted the rest of the time.
Table 1. Instruments of monetary policy
Notes: This table lists the names of all the quantitative instruments used by the Banque de France and specifies when they were first implemented (introduction) and when they were last used (abolition). Monnet (2014) shows details on the use and definition of the instruments as well as the changes in the value of each instrument from when it was introduced to the end of 1973.
There are two main reasons why a single instrument or a compound index of instruments cannot be used as a measure of monetary policy when, in post-war France ceilings and ratios -- rather than open market operations, the money base, or interest rates -- are the primary instruments of central bank policy.
- First, no single quantitative instrument was used – or kept the same definition – over the period. A combination of different instruments always had to be applied, and the particular choice of combination varied over time.
Direct bank credit controls had to be supplemented with various liquidity ratios (rediscount and credit ceilings) in order to be effective. For example, if a bank reached its rediscount ceiling, it could sell bonds or substitute demand deposits for time deposits or mid-term credit for short-term credit to increase its liquidity and its ability to lend. Liquidity ratios thus served to block these substitution effects.
- Second, and more importantly, even when one instrument was used over a long period, the values of that instrument over time are not commensurable. What matters is not the nominal value of the ratio or the ceiling, but whether it is constraining.
For example, an increase in the Treasury bill floor or in the liquid asset ratio is not a restrictive measure if, as in 1956 and 1962, it only served to keep pace with the changing composition of banks' balance sheets, without actually imposing a tighter constraint. This difficulty is compounded by the fact that exemptions were applied to certain instruments at different points in time, and that the combinations of instruments used changed over time.
Hence, it is essential to know the intentions of the policymakers observing bank balance sheets, the constraining effects of their instruments, and when the decision to change a ceiling or a ratio is taken.
Measuring the impact of a central bank’s actions
For this reason, I follow Romer and Romer (1989) and use narrative evidence to build a measure of the central bank’s actions, based on whether or not French central bankers intended to pursue a restrictive policy. I examine archival evidence directly concerning policymakers’ intentions and decisions, and I construct a dummy variable that is equal to 1 when quantitative tools are implemented by the Central Bank in order to fight inflation. The paper identifies six episodes of restrictive monetary policy (Table 2).
Table 2. Dates of restrictive monetary policy
Then, I estimate a VAR with monthly data where I simulate a shock on the dummy variable measuring the policy stance. I find that monetary policy shocks had a significant and sustained impact on production and the price level (Figure 1). When policy turned restrictive, industrial production, prices, and the money supply decreased by 5% within 20 months. Contrary to most VAR studies, there is no price puzzle. Variance decompositions show that monetary policy shocks in France explain approximately 40% of the variance in industrial production and price levels.
Figure 1. Impact of a restrictive monetary policy shock. VAR with 4 variables (Dummy, M2, CPI, Production).
Conversely, a shock to the discount rate in a VAR model does not produce significant or consistent responses in production and prices (Figure 2). In such an economy with regulated rates and ubiquitous quantitative controls, measuring the policy stance with an interest rate is clearly misleading.
Figure 2. Influence of a rise in the French discount rate. VAR with 4 variables (M2, CPI, Production, Banque de France discount rate).
First, this study elucidates the fact that monetary policy was neither absent nor passive during the early post-war period in Europe -- before the Great Inflation and under the Bretton Woods system – periods of fixed exchange rates and ubiquitous financial restraints (Eichengreen 2008).
Second, it shows that quantitative controls on money or credit can be effective in the short-term to decrease output and prices.
Third, it makes clear that the difficulty to measure the policy stance when quantitative controls are in place arises from the fact that the central bank has to change its instruments constantly to adapt to financial innovation and to the circumvention of previous instruments.
Fourth, it corroborates the case for a combination of VAR methodology and the ‘narrative approach’ (Romer & Romer 1989) to offer robust, stylised facts useful to the construction of business-cycle models.
What we learn from the French experience is that the well-known hump-shaped impulse response functions (Christiano et al. 1999) can be obtained without a liquidity effect. A decrease in quantities is not necessarily equivalent to an increase in prices. This is especially useful for assessing monetary policy in countries that used or still use quantitative controls as India or China.
Historical lessons for current macroprudential policies are not straightforward (Kelber and Monnet 2014). French 1950-60s policies were implemented against a backdrop of highly regulated and relatively closed national financial systems, where an excess supply of credit translated primarily into upwards inflationary pressure rather than into a banking or financial crisis. Credit controls were often backed up with capital controls and could thus be targeted specifically at the banking sector without leading to an exchange of assets with the rest of the financial sector. Thus, the objectives of credit controls and the state of the financial system were very different from the current situation in Europe. That said, history still demonstrates that the implementation of macroprudential tools can have a significant impact on monetary policy: The use of quantitative instruments to control credit and bank liquidity necessarily influenced money creation and inflation (especially with a low level of bank disintermediation), and often had an impact on credit allocation as banks substituted the assets on their balance sheets. For this reason, asset substitution was key to the functioning of the quantitative instruments and most of central banking debates at that time focused on the complementarities between the liquidity and credit controls.
Borio, C (2011), “Implementing a macroprudential framework: Blending boldness and realism”, Capitalism and Society, 6(1).
Christiano, L, M Eichenbaum, and C Evans (1999), “Monetary policy shocks: What have we learned and to what end?”, in Handbook of Macroeconomics, Vol.1, . p.65-148.
Eichengreen, B (2008), The European economy since 1945: coordinated capitalism and beyond, Princeton University Press.
Friedman, M and A Schwartz (1963), A monetary history of the United States, 1867-1960, Princeton University Press.
Galati, G, and M, Richild. (2013), “Macroprudential policy–a literature review”, Journal of Economic Surveys, 27(5), 846-878.
Kelber, A, and E Monnet (2014), “Macroprudential policy and quantitative instruments: a European historical perspective”, Financial Stability Review 18, p.151-160.
Monnet, E (2013), “Financing a planned economy. Institutions and credit allo¬cation in the French Golden age of growth (1954-1974)”, Berkeley Economic history Lab working paper n°2.
Monnet, E (2014), “Monetary policy without interest rates. Evidence from France’s Golden Age (1948-1973) using a narrative approach”, American Economic Journal: Macroeconomics, October, forthcoming.
Romer, C and D Romer (1989), “Does monetary policy matter? A new test in the spirit of Friedman and Schwartz”, in NBER Macroeconomics Annual, MIT Press.