In November 2008, at an event at the London School of Economics, Queen Elizabeth II asked economists together: "Why did not anyone notice it?" This meant the 2008 financial crisis, which could not be predicted in its entirety. This led to an ongoing discussion on the relevance of macroeconomic forecasts and the general question of the adequacy of the models used at that time. Critics have cited a number of causes responsible for the poor quality of forecasts.
The reasons are complex and generally relate to the underlying assumptions of the model. For example, many hierarchical models viewed the financial sector in a rudimentary way. Another criticism was that empirical relationships are often supposed to be constant, for example. between unemployment and inflation. During the financial crisis, these models observed in the past suddenly seemed useless.
The relevance of forecasts in the decision-making process of monetary policy
These and other issues were addressed in the macroeconomic discussions during and after the financial crisis. Empirical models have been extended accordingly and their predictive power improved in a sustainable manner. But why do economists have an interest in making macroeconomic forecasts? A striking example of the relevance of economic forecasts is their use in central banks.
The economic look
Every Monday, the "National Economic Society" (NOeG), in co – operation with the "Press", writes a blog on a current economic topic: NOeG is a non – profit association for the promotion of l & # 39; economy.
Contributions from external authors do not necessarily reflect the opinion of the "press" publisher.
Modern monetary policy is based on the game with the expectations of households and businesses regarding future developments in interest rates and prices. Central banks must analyze their monetary policy in a dynamic perspective. This means that the effect on variables such as unemployment and inflation, as well as on underlying expectations, must be measured. These effects are mainly taken into account in dynamic models. However, central banks must make the right decisions at critical moments. Thus, if the US Federal Reserve decides to lower interest rates, as at the July meeting, then this decision is based on short- and medium-term business cycle forecasts.
The predictive power of modern methods
The quality of monetary policy decisions is closely related to the forecaster's success rate. This raises the question of the accuracy of central bank forecasts and the ability of conventional methods to detect a crisis such as the Great Recession in a timely manner. For example, we examine differences in the quality of prognosis between traditional macroeconomic models and more recent methods, such as those currently used, for example. by the European Central Bank (ECB) or the US Federal Reserve.
The key question is whether modern models have been able to predict slowing down? To answer that, we realize a thought experiment. We expect to be in the third quarter of 2008. Most economic research institutes and central banks have already seen increased uncertainty in the financial markets. However, it was unclear to what extent they affected real variables such as economic growth or inflation. The next step is to use the data available until the third quarter of 2008 and to forecast inflation during the crisis. Forecasts are based on two models: first, an empirical standard model used by central banks since the late 1980s; and secondly, an improved model that can detect both changes in macroeconomic relationships and the efficient processing of large volumes of data. (eg from the financial sector).
The figure shows the predicted value of inflation (black line, dotted line) and the uncertainty of this one-time forecast (shaded areas, also called forecast intervals). It can be seen that the traditional model underestimates the immediacy of falling prices. In addition, forecasts are subject to considerable uncertainty. Another problem is that the realized value of inflation (black line) lies outside the shaded areas. This means that the underlying model considers that a sudden collapse of inflation is unlikely.
In contrast, the extended model provides a much more accurate inflation forecast. The risk of deflation is immediately indicated. This is true for the point forecast as well as for the forecast intervals, which predict the slowdown in inflation with a relatively high probability. During the crisis, this assessment would have provided a better basis for the ECB's monetary policy decisions.
Our example obviously benefits from the fact that, retrospectively, all the facts are known and that the evolution of macroeconomic fundamentals is observable. Nevertheless, the experience gained during and after the financial crisis has encouraged the development of more realistic models that have significantly increased the accuracy of macroeconomic forecasts. The question of whether the new methods will prove themselves in real-world conditions and in real time will become apparent during the next economic crisis.
Florian Huber (* 1987 in Spittal an der Drau) is Professor of Economics at the University of Salzburg.
Michael Pfarrhofer (* 1993 in Linz) is research assistant at the University of Salzburg and at the University of Economics and Commerce of Vienna.