In 1989, New Zealand’s central bank moved to a new method for controlling inflation that has since become known as ‘inflation targeting’. In subsequent decades, much of the rest of the world adopted this regime, under which central banks commit to a target for inflation and then meet it by manipulating their policy instruments, particularly the short-term interest rate, to counter the effects of shocks hitting the economy. For instance, if positive aggregate demand shocks raise inflation above target and output above potential, the central bank raises policy interest rates to curtail spending and thereby bring output and inflation back towards their targets.
It was recognised that other types of shocks – such as aggregate supply shocks and financial sector and external sector shocks – would prove challenging for the framework. By lowering output below potential but raising inflation above the target, aggregate supply shocks put the central bank in a quandary about the direction in which to move its instrument. lowering interest rates, for example, would help with the first goal but not the second. Avoidance of financial sector and external sector crises, which was often an additional goal for central banks, was also difficult to achieve through mere manipulation of interest rates, but, for a long time, inflation targeting worked well in practice, even if not in theory. Over time, central banks grew nimbler at balancing output and price goals in response to aggregate supply shocks until such time as the supply shocks reversed themselves; in essence, central banks relied on the transitory nature of supply disturbances as an additional instrument. Likewise, as documented in this column, central banks turned to macroprudential policies as the first line of defence against asset price inflation (particularly house price inflation), leaving interest rates free to pursue the macro goals of price and output stabilisation (see De Grauwe 2007). Central banks in emerging markets and developing economies (EMDEs) used foreign exchange intervention as an additional instrument alongside interest rates to achieve their goals of taming exchange rate volatility and enduring adequacy of foreign exchange reserves (see Ghosh et al. 2012 for a clear discussion of this strategy).
The long-lived disruption of global supply chains during the pandemic led to a widespread upsurge in inflation in 2021-22, denting the hitherto sterling reputation of inflation targeting (IT) regimes. As inflation ran above targets in many economies in 2021-22, previous concerns regarding the resilience of inflation targeting regimes resurfaced. As early as 2012, Frankel (2012) had expressed reservations regarding the ability of inflation targeting to contain inflation in the face ‘of asset bubbles and supply side shocks’. He went so far as to pen an obituary for inflation targeting regimes and suggest that they should be replaced by nominal GDP targeting. In a similar vein, McKibbin and Panton (2018) argued that while “inflation targeting has appeared to work well in Australia in the past decades, the likely importance of supply shocks in the future – due to, developments such as geopolitical fragmentation or climate events – suggests that nominal income targeting is worth considering as an evolutionary change to Australia’s framework for monetary policy”. Leijonhufvud (2007) expressed concerns about the “troubling legacy of asset price inflation” arising from inflation targeting’s elevation of macro goals over financial stability goals.
Has inflation targeting outlived its usefulness?
In Bhalla et al. (2023), we carry out a comprehensive examination of the performance of inflation targeting across advanced and emerging market economies since 1989. Our overall assessment of the regime is favourable but hedged: we argue that, while inflation targeting has worked well in the face of myriad challenges, credit for how much inflation targeting contributed to the pre-2021 decades of low inflation has run ahead of the evidence.
Inflation targeting started out with a bang. All nine of the initial adopters of the regime saw a decline in the three-year average inflation following adoption.
Figure 1 Inflation targeting’s initial success
Two quick wins
This successful initial policy offering was followed by two ‘quick wins’ as a couple of initial fears about the regime proved to be unfounded. The first fear had been that emerging markets lacked the human and intellectual capital to adopt inflation targeting regimes and that it would thus remain an advanced economy curiosum (e.g. Masson et al. 1997); however, over the past two decades, many emerging markets have successfully adopted the regime, making it a universal standard.
The second fear had been that central banks would become – in the colourful phrase of Mervyn King (1997) – “inflation nutters” (that is, they would elevate achievement of the inflation goal above all others). In practice, central banks have implemented inflation targeting in a flexible manner, giving weight to both output stabilisation and inflation control. For instance, Dąbrowski et al. (2025) study 36 inflation targeting frameworks across advanced economies and EMDEs and find most countries practice flexible inflation targeting.
But, despite the initial success and these two quick wins (universal adoption and flexibility in implementation), our overall assessment of inflation targeting is hedged: we think that credit for how much inflation targeting contributed to low inflation has run ahead of the evidence. In our paper, we provide evidence that:
1. The decline in inflation under IT regimes may simply reflect ‘regression to the mean’. The Ball and Sheridan (2004) critique states that inflation may have gone up for transitory reasons in late 1980s and early 1990s and would likely have come down even in the absence of IT adoption. Inflation targeting is merely a placebo that gets undeserved credit. Ball and Sheridan offer a telling analogy to make their point. Batting slumps – temporary phases where batters have trouble getting hits – are a ubiquitous feature of baseball. Suppose that, during one such slump, a batter is advised by his coach to sleep with his bat next to his pillow. The slump disappears, as slumps do, but the batter incorrectly gives credit to his coach’s advice. Proponents of inflation targeting regress the average change in inflation rate on a dummy variable for IT adoption and show that the coefficient estimate is negative. However, this coefficient becomes insignificant once regression to the mean is controlled for by including the pre-IT inflation rate as an additional explanatory variable.
We illustrate this by comparing the change in inflation rate after adoption of inflation targeting with the initial inflation rate in Figure 2. Countries that experienced a higher inflation rate (shown on the x-axis) happened to have also been the ones which experienced the largest declines in their inflation (y-axis) after the adoption of inflation targeting. We find that Ball and Sheridan’s conclusion still holds, even with a few additional decades of data; moreover, it holds with as much force for EMDEs as for advanced economies. In short, ‘regression to the mean’ remains a powerful explanation for the supposed better inflation performance of IT adopters.
Figure 2 Initial inflation and decline in inflation: Regression to the mean
2. Some, such as Rogoff (2003), have argued that alternative factors such as globalisation may be behind the fall in inflation, but credit is often given to inflation target adoption. Prima facie evidence for Rogoff’s point is presented in Figure 3, which shows median inflation across advanced economies and EMDEs for IT and non-IT countries. It is evident that some factors were contributing to a global rise in inflation in the 1990s and since then inflation has moderated across both advanced economies and EMDEs, and this was irrespective of whether the country adopted a formal inflation target regime.
Figure 3 Median inflation
We go further in trying to isolate the contribution of inflation targeting to moderating inflation through a country-by-country synthetic control method (SCM) analysis. That is, we compare the inflation performance in an inflation-target-adopting country with that of a synthetic ‘twin’ country that did not adopt an inflation target regime. The synthetic ‘control’ country shares the key characteristics of the ‘treatment’ country (the inflation target adopter), including similar inflation experience prior to the adoption of an inflation target. The results from our synthetic control analysis suggest that inflation targeting significantly lowered inflation (compared to that in the synthetic control) in only six of our 23 cases. This is shown in light blue in Figure 4.
Figure 4 Inflation performance: Inflation target adopters versus synthetic controls
Note: Light blue colour signifies statistical significance at 95% confidence level
Conclusion
Despite chinks in its theoretical armour, inflation regimes have performed well in practice partly due to the flexibility of the framework. Faced with numerous shocks such as financial crisis or supply disruptions, central bankers have expanded their policy toolkit to manage often conflicting objectives. A decade after Reichlin and Baldwin (2013) summarised the debate on inflation targeting, we concur that inflation targeting must evolve to survive and maintain its credibility. At the same time, we caution against attributing the great moderation in inflation to the adoption of inflation targeting – alternative explanations such as the effects of globalisation or regression to the mean should be given serious consideration.
Source: cepr.org
