Forecasting is possible

“Based on our regular economic and monetary analyses, we today conducted a comprehensive assessment of the economic and inflation outlook and our monetary policy stance.”

(Mario Draghi)

During the press conference of 8th December 2016, the President of the ECB (European Central Bank) Mario Draghi, together with the Vice-president Vítor Constâncio, presented a series of analyses on the current economic situation, the most likely outcome of the next few years and the actions they are going to take.

Not focusing on the main news of the press conference, that is the extension of the Asset Purchase Programme (APP) at the current monthly pace of 80€ billion until the end of March 2017, I will focus instead on the forecasts provided by the ECB regarding inflation and GDP growth. Specifically, the ECB has forecast an increase in the annual level of inflation in the next three years, from an annualized value of 0.2% in 2016 to an annualized value of 1.7% in 2019 (still being below the target of 2%). It has also forecast a constant annual real GDP growth, being 1.7% in 2016 and 2017 while being 1.6% in 2018. A summary is provided in the picture below.

 

 

In this article I will try to reproduce the forecasts of the ECB with econometrics tools. Both data on inflation and GDP growth are taken from the site of ECB.

Inflation

Let’s start with the inflation. The following graph represents the time series of the monthly values of inflation in the EU area from January 1999 to November 2016.

After some controls on the properties of the time series, I have modelled it as an Autoregressive process of order 1, using jargon an AR (1). The choice of the order was carried out by means of statistical methods and will not be discussed in details in this article. According to this specification of the time series of inflation I have tried to forecast its values in the next three years. A graphical representation follows.

The red line is the actual time series, while the blue line is the forecast. From December 2016 till November 2019 the blue line is surrounded by a grey area being the confidence interval, namely the area in which the fitted values might actually take value. Even if the confidence interval is quite big and it reduces the precision of the analysis, it is possible to observe that inflation is expected to grow, reaching a value of 0.93% in November 2017, of 1,18% in November 2018 and of 1.3% in November 2019.

Comparing this analysis with the forecast of the ECB, the latter seems to overestimate the expected annualized inflation in the next three years. However, this is due to a better and more complicated specification of dynamics of future values on inflation, since here I am constructing a univariate model, not including all the kind of variables that might affect future values of inflation. Still, the trend of expected annualized inflation is increasing in both analyses.

GDP growth

Let’s move to the GDP growth. The following graph represents the time series of the quarterly values of GDP growth in the EU area, namely first-differences, expressed in percentage changes from the first quarter of 1995 to the third quarter of 2016.

Again, after some controls on the properties of the time series, I have modelled it as an Autoregressive Moving Average process of order 1, ARMA (1,0). According to this specification of the time series of GDP growth in percentage changes I have tried to forecast its value in the next three years. A graphical representation follows.

Again, the red line is the actual time series while the blue line is the forecast. The confidence interval is not as big than in the inflation case, and it’s possible to observe that quarterly GDP growth is expected to remain constant having the value of 0.68% in the third quarter of 2017, of 0.69% in the third quarter of 2017 and of 0.7% in the third quarter of 2019. Moving to an annual estimate, it would mean a constant annual GDP growth of 2.8%, one percentage point more than the estimate of the ECB. Nonetheless, the trend of expected GDP growth is expected to remain costant in both analyses.

As I said before, there is no doubt that the estimates communicated by the ECB are more precise and accurate that the ones presented in this article. Still, its aim was to show that it might be possible to reproduce a similar forecast on main macroeconomic variables such as inflation and GDP. For the sake of brevity, I have left away the details of the specification of both analyses, but I would gladly explain them for the most curious ones.

One thought on “Forecasting is possible

Leave a Reply