The ability of the sentiment indicators to forecast stock market crashes
This study aims to describe and quantify the short-term relationship between the sentiment indicators and the aggregated indicators of the real economy.
The novelty of this paper comes from the following two directions: (i) first of all we tried to improve the existing theoretical and empirical evidence by using a set of methodologies for measuring the investors’ sentiment, and, in order to have a more accurate outlook on the development and regulations’ conformity degree of a country, we constructed a composite indicator of different real economy variables for the economy’ stance; (ii) then, we tested the ability of investors’ sentiment to forecast assets’ prices not only during normal market conditions, but also under extreme distributions.
Data and methodology
We used 2000-up to present quarterly observations for the EU-27 countries in order to study the causal relationship between investors’ perception on the economy and the indicators of the real economy. An out-of-sample forecasting test was conducted in order to analyze the ability of investors’ sentiment indicators to predict the evolutions of the real economy, especially during the financial crisis.
As methodologies, we opted to test whether the stock return Granger causes the consumer confidence on a very short term and if the relationship is reciprocal. Nonetheless, in order to isolate the key-determinants of the financial crises we employed a logit model using as (a) explanatory variables the following real economy’ variables: stock return, real GDP growth, market liquidity, inflation, interest rate, share of domestic lending in GDP, sovereign debt, budgetary deficit, FX rate, monetary aggregate, but also, for a better grasp of a macroeconomic view, we constructed a special composite crisis indicator, and as (b) behavioral variables: European sentiment indicator (chosen as to overcome the shortcoming indulged by the fact that the national stock markets of the European countries are closely integrated) and the consumer confidence index (as alternative proxy for investors’ sentiment).
Expected results
Our model rests on two sets of assumptions- one about market frictions (we assumed that there are short-sales constraints) and the other about investor behavior (with respect to the latter, we posited the existence of a class of irrationally overconfident investors, where overconfidence was seen as a tendency to overestimate the relative precision of one’s own private signals).
The preliminary results show that the aggregate measures and investors’ sentiment are highly correlated and that in a multiple regression, both have incremental predictive power for the future market returns and also for the tendencies of the real economic activity.