Kenza Benhima & Céline Poilly, 2021, Journal of Monetary Economics, vol. 117, 278-295
It is commonly accepted that expectations on economic activity can be important drivers of fluctuations by generating waves of optimism and pessimism. The recent events of the last decades –such as the Great Recession or the Covid crisis – are stimulating interest in analyzing the role of misperceptions in the business cycle. Misperception means that agents are badly informed about the state of the economy, receiving noisy signals. Survey expectation errors usually help to identify misperception shocks. When professional forecasters over-estimate output growth, this may reflect excessive optimism either about the state of supply or about the state of demand. A large part of the literature has focused on misperception of total factor productivity, called supply-driven noise shocks. It can be assumed that agents also receive noisy signals about aggregate demand. How can misperceptions about supply and demand be disentangled? We show in this paper that expectation errors regarding output growth and inflation are key information needed to identify supply-driven and demand-driven noise shocks and to quantify their contribution to the cycle.
This paper builds a reduced-form model composed of decision makers (households and firms) and survey participants (“nowcasters”) to capture the effects of noise (misperception) shocks on output, inflation and the associated expectation errors. Importantly, decision makers are better informed than nowcasters, although they all receive a noisy signal about supply and demand. This information structure rationalizes the now common practice in the literature of using professionals’ expectations to uncover noise shocks. This stylized theoretical model provides key identification restrictions for the estimation of demand and supply noise shocks in the data: If demand (supply) shocks drive a positive (negative) correlation between output and inflation, they should drive a positive (negative) correlation between errors about output and about inflation.
Figure 1. Nowcast error of real GDP growth and GDP deflator inflation in percentage points over 1968q1-2017q1.
Based on these predictions, it is possible to disentangle demand and supply noise shocks in the data. We estimate a multivariate empirical model – namely a Structural Vectoral AutoRegressive model – on US data over the sample 1969q1–2017q1, including real GDP growth, the inflation rate and their corresponding nowcast errors. Shocks are identified using the sign restrictions suggested by the reduced form model. We find that demand noise shocks explain 24% of output growth volatility, while supply noise shocks explain only 8%. Imperfect information on demand therefore has deeper consequences on economic activity than imperfect information on supply.
Using a micro-founded New Keynesian model, we argue that the effect of demand noise shocks on output originates in a “monetary policy channel”. As the central bank receives a positive signal about demand, it expects a rise in inflation and increases the interest rate accordingly. This rise in interest rate then depresses aggregate demand, leading to recessionary demand noise shocks. This channel seems to be operational in the data, as we observe a rise in the nominal interest rate in response to a positive noise demand shock, i.e. higher misperception of demand.
This study opens up new avenues for future research. The central role that monetary policy seems to play in the transmission of noise shocks calls for further investigations of central banks’ information. This line of research could provide valuable input to the design of optimal policies that are conditional on both private agents’ and policy makers’ imperfect information.
→ This article was issued in AMSE Newletter, Winter 2021.