Macroeconomic outcomes in Britain’s interwar years were terrible – they featured two of modern Britain’s worst recessions, unemployment twice peaked above 20% and was rarely below 10% and there were two periods of chronic deflation. Policy, meanwhile, was pulled in multiple directions by multiple objectives – employment, price and financial stability and debt sustainability. These challenges gave birth to modern macroeconomics, inspiring the work of John Maynard Keynes. In a new working paper, I apply modern empirical techniques to look at the period with fresh eyes. I find that monetary and fiscal policy played a central role in macroeconomic developments – and that outcomes could have been better had policymakers been less wedded to the traditional policy consensus, and especially the Gold Standard.
Asking what role macroeconomic policy played in interwar Britain is not a new topic, so why revisit it? First, although there have been some excellent recent attempts to answer narrow questions about the period (on tax, defence spending, uncertainty, inflation expectations, protectionism and exit from the Great Depression), the most recent comprehensive assessment of macroeconomic policy dates to 1995. Second, and more generally, many of the key results of modern empirical macroeconomics rest on one period and one country, the post-war United States, which contains a relatively limited set of events to exploit (Ramey (2016) offers a nice overview).
What do I do?
In the paper, I build an empirical model of the relationships between the economy and policy variables and specify how deviations from those relationship should be interpreted in terms of ‘fundamental’ shocks – in more technical terms, I estimate a vector auto-regression (VAR) model, structurally identified with sign restrictions, taking inspiration from Mountford and Uhlig (2009). This approach is both flexible, allowing for a wide range of ways to think about shocks (including contemporary responses, which traditional recursive approaches preclude), and can accommodate all my shocks of interest – two business cycle shocks (demand and supply) and three policy shocks (monetary policy, tax and government spending). I define the policy shocks so that each of a rise in Bank Rate, increase in taxes or cut in spending raises unemployment and reduces prices; a demand shock sees prices fall as unemployment rises, while a supply shock sees prices rise as unemployment rises. To overcome recent critiques of some aspects of Mountford and Uhlig’s methodology, I use the estimation approach suggested by Arias et al (2018), as implemented in the fabulous BEAR Matlab toolbox.
All of my data is at monthly frequency and is taken from sources available at the time. The baseline model is a VAR containing unemployment, the price level, Bank Rate, tax receipts and government spending. The data on the public finances are a particularly important innovation: I hand-collected them from the Government’s official journal, The Gazette – and to my knowledge they have not been used elsewhere at this high frequency (though a quarterly version has been used to look at tax and uncertainty).
What do I find?
Chart 1: Response of unemployment and prices to five shocks
Notes: Charts show the change in each variable in response to a 1 standard deviation shock. Shaded areas show the 68% credible set. Units are per cent for inflation and percentage points for unemployment.
Chart 1 neatly summarises my core results. It plots how prices and unemployment react to each of my shocks. For example, the first column shows that a typical negative demand shock increases unemployment by a peak 0.25 percentage points and reduces prices by more than 0.5% at its largest effect. These shocks are symmetrical by design, but for the sake of comparison are all presented here such that they lead to a rise in unemployment. Strikingly, all three of the policy shocks – monetary policy, spending and tax – affect unemployment by roughly the same amount, at least initially. But after three years (ie 36 months), while the effects of spending and tax shocks have largely faded, those of monetary policy persist. The effects on prices are somewhat more varied: monetary policy pushes down on the price level for an extended period; spending shocks push down by about as much, but the effect starts to unwind after about a year, while; tax shocks have a milder but more persistent downward effect on the price level.
To draw comparisons against the wider literature I re-estimate the baseline model with GDP in the place of unemployment. I find that the share of variation in GDP accounted for by my policy shocks is high relative to other estimates – 8.6%, 12% and 9.2% after two years, for monetary, tax and spending shocks respectively, compared to reported ranges of 0.5%–8.8%, 0.5%–4.8% and 2.9%–12.6% respectively. Similarly, my estimates of fiscal ‘multipliers’ (roughly, the change in GDP for a given change in tax or spending) are towards the upper end of the wider literature, at 4¾ for tax and 1½ for spending, compared to a reported range of 1 to 5 for tax and 0.2 to 2 for spending. These estimates are also large relative to era-specific estimates – Cloyne et al (2018) estimate a tax multiplier of 2.3 for the interwar years, while Crafts and Mills (2013) estimate a (defence) spending multiplier of 0.3 to 0.8 from the rearmament ahead of World War II. So policy – monetary and fiscal – could have powerful effects in this period.
Chart 2: Historical decomposition of prices and unemployment over five periods
Notes: Solid black lines indicate the total contribution of shocks to the deviation of the variable from the model steady state at each point in time, with the coloured areas denoting the contribution from each shock. The price level is expressed as the percentage deviation, while unemployment is the percentage point deviation.
But my results go further than just showing that policy was powerful – the outcomes in the period were also driven by the decisions policymakers made. Chart 2 is perhaps the most eloquent summary of my full results, plotting the contribution to the price level and unemployment over the period of each of my five shocks. Taking each of the five periods in turn, the postwar ‘boom’ was clearly fuelled predominantly by the continuation of wartime levels of spending. The subsequent ‘bust’ was the product of the fiscal tightening that took place from 1920, exacerbated by tighter monetary policy and the coal strike of 1921. The weak expansion of the later 1920s, called ‘The Doldrums’ by Arthur Pigou, was supported in its middle years by lower taxes and looser monetary policy, though both were unwound towards the end of the decade as part of the restoration of the Gold Standard. ‘The Slump’ (as the Great Depression has traditionally been known in the UK) was significantly worsened by the tightening of monetary and fiscal policy mounted in defence of sterling – an attempt which ultimately failed. With Britain off the Gold Standard from September 1931, monetary policy was eased to 2%. It remained there for the rest of the period – called the ‘Cheap Money’ era – and made a material contribution to the steady expansion of the 1930s. Strikingly, although government spending was stepped up in the later 1930s to support rearmament, consistent with results elsewhere, I find that this made only a modest contribution to lowering unemployment.
What are the policy implications?
My results suggest that changes in fiscal policies could have had material effects on unemployment, but that these effects dissipated relatively quickly – so while countercyclical (or just less procyclical) policy would have improved outcomes, Keynes’ proposal of a sustained fiscal loosening would probably not have solved the problem of interwar unemployment. By contrast, although the peak effect of monetary policy was similar in scale to that of fiscal policy, its effects were more persistent – suggesting looser monetary policy, especially before the ‘Cheap Money’ era, would have been a more powerful way to address unemployment.
The implication of this is that better outcomes might have been achieved with alternative polices: a slower fiscal consolidation in the 1920s, a later return to gold (perhaps at a devalued parity) and a less aggressive defence of sterling in 1931 could all have kept unemployment lower and prices higher. Similarly, monetary and fiscal policy could have been more complementary, with somewhat looser monetary policy facilitating a slower fiscal consolidation. But doing this would have required agreement on pursuing different objectives – a weaker commitment to gold and balanced budgets – something that would have been challenging, given the centrality of the Gold Standard in policymakers’ understanding of Britain’s economic stability and global role.
David Ronicle is on secondment from the Bank’s Monetary Analysis department to the UK delegation to the IMF.
If you want to get in touch, please email us at firstname.lastname@example.org or leave a comment below.
Comments will only appear once approved by a moderator, and are only published where a full name is supplied. Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.