Skip to main content

Is forecasting inflation easier under inflation targeting?


This paper investigates whether monetary-policy regime changes affect the success of forecasting inflation. The forecasting performances of some linear and nonlinear univariate models are analyzed for 14 different countries that have adopted inflation-targeting (IT) monetary regimes at some point in their economic history. The results show that forecasting performance is generally superior under an IT monetary regime compared to nonIT (NIT) periods. In more than half of the countries covered in this study, superior forecasting accuracy can be achieved in IT periods regardless of the model used. In contrast, among most of the remaining countries, the results remain ambiguous, and the evidence on the superiority of NIT is limited to very few countries.

This is a preview of subscription content, access via your institution.

Fig. 1


  1. In addition to the special role given to IT, the prominence of inflation forecasting has been raised by the recent formalization of the New Keynesian optimal policy. The New Keynesian model has been used to demonstrate that the optimal choice of policy will depend on the optimal forecasts (see Svensson 2005; Faust and Wright 2012).

  2. See Brito and Bystedt (2010) for a counter argument that claims that there is no evidence that an IT improves economic performance, as measured by the behavior of inflation and output growth in developing countries.

  3. This interpretation is not directly in contrast with D’Agostino and Surico’s results, which are mentioned above. These results provide evidence that a policy regime that successfully stabilizes inflation in the US makes it harder to improve upon the forecasts that are based on “naïve” models. However, the evidence that we provide here can be interpreted thusly: a policy regime that successfully stabilizes inflation (i.e. an IT regime) makes it easier to forecast inflation irrespective of the underlying model that is used for forecasting.

  4. The effect of IT regimes on inflation volatility will be studied in future research.

  5. Deciding whether the direct or the iterated approach is better is an empirical matter because it involves a trade off between the estimation efficiency and the robustness-to-model misspecification; see Elliott and Timmermann (2008). Marcellino et al. (2006) address these points empirically using a dataset of 170 US monthly macroeconomic time series. They find that the iterated approach generates the lowest MSE-values, particularly if lengthy lags of the variables are included in the forecasting models and if the forecast horizon is long.

  6. This process involves replacing \(y_{t}\) with \(y_{t+h}\) on the left-hand side of Eq. (4) and running the regression using data up to time \(t\) to fitted values for corresponding forecasts.

  7. Indeed, \(d_{t}\) is convex in \(y_{t-1}\) whenever \(y_{t-1}<c\) and \(-d_{t}\) is convex whenever \(y_{t-1}>c\). Therefore, by Jensen’s inequality, naive estimation underestimates \(d_{t}\) if \(y_{t-1}<c\), and it overestimates \(d_t\) if \( y_{t-1}>c\).

  8. A detailed exposition of approaches for forecasting from a SETAR model can be found in van Dijk et al. (2003).

  9. See Franses and Dijk (2000) for a review of feed-forward-type neural network models.

  10. For the sake of brevity, we only provide the results of the iterative forecasts. The results that were obtained with direct forecasts are qualitatively similar and available upon request.

  11. As long as we assume the same variance for both periods, the DM test is still valid in this case. However, one may object to this assumption by indicating that IT can reduce the variance of the inflation.

  12. Monthly inflation forecasts are scaled by 100. As a caveat, one should keep in mind that comparing two MSE series via Diebold-Mariano statistics is a scale-dependent process, i.e. the statistics change under the multiplication of two series by a constant. Here, we scale the monthly inflation figures by 100 to express them in terms of monthly percentages. See Clark and West (2006) for a detailed discussion on the effects of scaling the out-of-sample MSE-based tests.

  13. To find out whether any specific model particularly performs better or worse than the others in IT or NIT periods, we have tested the relative forecast accuracy of the utilized models in each period against the benchmark of the random walk model. The null hypothesis was that the forecasts of the model under consideration are no better than those of the random walk model; the alternative is that random walk forecasts more accurately than the model under consideration. The associated DM test statistics indicated mixed results that vary across countries and forecasting horizons. Hence, we cannot assert that forecasts from a particular model are superior to random walk forecasts in most countries and horizons. Furthermore, comparing the results in IT and NIT periods did not lead us to conclude that one or more models overwhelmingly provide better forecasts in one period in comparison to other periods. Therefore, to save some space, we do not report these results here, which can be found in the working paper version and also available upon request.

    Table 3 The p-values of the DM tests where the null hypothesis is that IT forecasts are no better than NIT forecasts


  • Al-Qassam MS, Lane JA (1989) Forecasting exponential autoregressive models of order 1. J Time Ser Anal 91(2):95–113

    Article  Google Scholar 

  • Ball LM, Sheridan N (2004) Does inflation targeting matter? In: The inflation-targeting debate. NBER Chaps, National Bureau of Economic Research Inc., Cambridge, pp 249–282

  • Brito RD, Bystedt B (2010) Inflation targeting in emerging economies: panel evidence. J Dev Econ 91(2):198–210

    Article  Google Scholar 

  • Clark TE, West KD (2006) Using out-of-sample mean squared prediction errors to test the martingale difference hypothesis. J Econom 135(1–2):155–186

    Article  Google Scholar 

  • D’Agostino A, Surico P (2012) A century of inflation forecasts. Rev Econ Stat 94(4):1097–1106

    Article  Google Scholar 

  • D’Agostino A, Giannone D, Surico P (2006) (Un)Predictability and macroeconomic stability. Working Paper Series 605, European Central Bank

  • D’Agostino A, Gambetti L, Giannone D (2011) Macroeconomic forecasting and structural change. J Appl Econom 26(7):82–101

    Google Scholar 

  • De Gooijer JG, De Bruin PT (1998) On forecasting SETAR processes. Stat Probab Lett 37(1):7–14

    Article  Google Scholar 

  • Diebold FX, Mariano RS (1995) Comparing predictive accuracy. J Bus Econ Stat 13(3):253–263

    Google Scholar 

  • de Mendonça HF, de Guimarães e Souza GJ (2012) Is inflation targeting a good remedy to control inflation? J Dev Econ 98(2):178–191

    Article  Google Scholar 

  • Elliott G, Timmermann A (2008) Economic forecasting. J Econ Lit 46(1):3–56

    Article  Google Scholar 

  • Faust J, Wright JH (2012) Forecasting inflation, unpublished manuscript

  • Franses P, Dijk D (2000) Nonlinear time series models in empirical finance. Cambridge University Press, Cambridge

    Book  Google Scholar 

  • Gonçalves CES, Salles JaM (2008) Inflation targeting in emerging economies: What do the data say? J Dev Econ 85(1–2):312–318

    Article  Google Scholar 

  • Hamilton J (1994) Time series analysis. Princeton University Press, Princeton

    Google Scholar 

  • Kock AB, Teräsvirta T (2011) Forecasting with nonlinear time series models. In: Clements MP, Hendry DF (eds) Oxford handbook on economic forecasting. Oxford University Press, Oxford, pp 61–87

    Google Scholar 

  • Lin JL, Granger CWJ (1994) Forecasting from non-linear models in practice. J Forecast 13:1–9

    Article  Google Scholar 

  • Lin S, Ye H (2007) Does inflation targeting really make a difference? Evaluating the treatment effect of inflation targeting in seven industrial countries. J Monet Econ 54(8):2521–2533

    Article  Google Scholar 

  • Marcellino M, Stock JH, Watson MW (2006) A comparison of direct and iterated multistep AR methods for forecasting macroeconomic time series. J Economet 135(1–2):499–526

    Article  Google Scholar 

  • Rossi B, Sekhposyan T (2010) Have economic models’ forecasting performance for US output growth and inflation changed over time, and when? Int J Forecast 26(4):808–835

    Article  Google Scholar 

  • Stock JH, Watson MW (2007) Why has US inflation become harder to forecast? J Money Credit Banking 39(1):3–33

    Article  Google Scholar 

  • Stock JH, Watson MW (2009) Phillips curve inflation forecasts. In: Fuhrer J, Kodrzycki YK, Little J, Olivei GP (eds) Understanding inflation and the implications for monetary policy: a Phillips curve retrospective. The MIT Press, Cambridge

    Google Scholar 

  • Svensson LE (2010) Inflation targeting. In: Friedman BM, Woodford M (eds) Handbook of monetary economics, handbook of monetary economics, vol 3, 22nd edn. Elsevier, Amsterdam, pp 1237–1302

    Google Scholar 

  • Svensson LEO (2005) Monetary policy with judgment: forecast targeting. Int J Cent Bank 1(1):1–54

    Google Scholar 

  • Teräsvirta T (2006) Forecasting economic variables with nonlinear models. In: Elliott G, Granger C, Timmermann A (eds) Handbook of economic forecasting, handbook of economic forecasting, vol 1, 8th edn. Elsevier, Cambridge, pp 413–457

    Chapter  Google Scholar 

  • van Dijk D, Franses PH, Clements MP, Smith J (2003) On SETAR non-linearity and forecasting. J Forecast 22(5):359–375

    Article  Google Scholar 

  • White H (2006) Approximate nonlinear forecasting methods. In: Elliott G, Granger C, Timmermann A (eds) Handbook of economic forecasting, vol 1, 9th edn. Elsevier, Cambridge, pp 459–512

    Chapter  Google Scholar 

Download references

Author information

Authors and Affiliations


Corresponding author

Correspondence to Harun Özkan.

Appendix: A note on Monte Carlo numerical method

Appendix: A note on Monte Carlo numerical method

As mentioned in the text, a version of Monte Carlo approach, which was first suggested by Lin and Granger (1994), is adopted for numerical computation of the multistep iterative forecasts within LSTAR and ARNN models. The main reason behind this choice is computational speed and accuracy of Monte Carlo simulation against the alternative approach of numerical integration: as the forecasting steps get higher, numerical integration becomes significantly slower.

Computing more than one step forecasts via Monte Carlo framework for nonlinear models in general consists of the following steps:

  • Step 1: Compute \(\hat{y}_{t+1|t}\) by directly plugging in \(y_t, y_{t-1},\ldots \) into the estimated equation.

  • Step 2: Generate \(n\) normal random variates with a mean of zero and a variance of \(\hat{\sigma }^2\) to form a vector of simulated \({\varepsilon }_{t+1|t}\) values.

  • Step 3: Compute simulated \(y_{t+2|t}\)s by plugging in the simulated values of \({\varepsilon }_{t+1|t}\) along with \(y_{t+1|t}\) and \(y_t, y_{t-1},\ldots \, n-\)times.

  • Step 4: Compute the Monte Carlo estimation of \({y}_{t+2|t}\) which is \(\hat{y}_{t+2|t}\).

  • Step 5: Repeat Steps 2, 3, and 4 to increase \( t \) for getting the higher step forecasts until the end of the forecast horizon.

Notice that, in order to apply a Monte Carlo scheme for forecasting it is necessary to assume a probability distribution for the error terms \( \{ \varepsilon _{t} \} \). Here, in all models {\({\varepsilon }_{t}\}\) is assumed to be i.i.d. \(N(0, {\sigma }^2)\) for all \(t\).

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Özkan, H., Yazgan, M.E. Is forecasting inflation easier under inflation targeting?. Empir Econ 48, 609–626 (2015).

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI:


  • Inflation targeting
  • Forecasting inflation
  • Forecast accuracy

JEL Classification

  • C45
  • C53
  • E31
  • E37
  • E42
  • E47
  • E52
  • E61
  • E65