Research

publications

“Forecasting housing investment” (with Gabe De Bondt and Arne Gieseck)

Journal of Forecasting, vol. 42, Issue 3, Special Issue: Advances in Forecasting in Macroeconomics and Financial Markets, April 2023, pp. 543-565. LINK

ECB, Working Paper Series, no. 2023/2807, April 2023. LINK Presented at the ESCB WGF (Working Group on Forecasting).

Abstract: This study applies a model averaging approach to conditionally forecast housing investment in the largest Euro area countries and the Euro area. To account for substantial modeling uncertainty, it estimates a large and diverse number of vector error correction models using a wide set of long- and short-run determinants and applies subset selection based on in-sample and out-of-sample criteria. First, a pseudo out-of-sample forecast exercise shows that our model averaging approach consistently beats a battery of distinguished benchmark models, including BVARs, FAVARs, LASSO, and Ridge regressions. This evidences that model averaging provides more accurate forecasts also in the case of housing investment. Second, we find remarkable cross-country heterogeneity in the drivers of housing investment. Overall, these findings guide forecasters and modelers on improving housing investment models and policymakers on implementing country-specific housing market policies.

working papers

“Leaning against housing booms fueled by credit”

National Bank of Slovakia, Working Paper Series, n. 9/2023. LINK

Review & Resubmit at Journal of Financial Stability.

Abstract: This study aims to empirically identify the state of the US housing market and establish a countercyclical state-dependent macroprudential policy rule. I do so by estimating a Markov switching model of housing prices, in which mortgage debt affects house prices nonlinearly and drives state transition probabilities. Second, I propose a state-contingent policy rule fed with the probability of being in each state, which I apply to setting a housing countercyclical capital buffer, a mortgage interest deduction, and a dividend payout restriction. Finally, I show that such hypothetical tools contain early warning information in a forecasting exercise to predict the charge-off rates of real estate residential loans and a financial stress index. The significance of this study is that it informs policymakers about the state of the housing market mechanically, while also providing a general rule to implement a state-contingent and timely macroprudential policy.

work in progress

“The effects of uncertainty shocks in the euro area” (with Arne Gieseck)

Conditionally accepted at the National Bank of Slovakia WPS, presented at the ESCB WGF. Latest version: LINK

Abstract: This paper estimates the effects of uncertainty shocks on economic activity in the euro area. We take the following approach to carefully identify the effects of uncertainty. First, we build a large macro dataset with euro area wide data which we summarize by principal components. Second, we estimate a FAVAR model using four prominent measures of uncertainty and our large dataset. Third, we identify an uncertainty shock by imposing sign and narrative sign restrictions, borrowing from standard uncertainty literature and from recent monetary policy literature on information shocks, respectively. We find that uncertainty has a significantly negative effect on economic activity measures in the euro area, while having a muted effect on savings.

“Non-linear effects of standard monetary policy shocks on housing: Evidence from a CESEE country” (with Adriana Lojschová and Alicia Aguilar)

Accepted at the ESCB ChaMP Research Network. Presentation: LINK

Abstract: This study estimates the effects of standard monetary policy shocks on house prices and housing investment in Slovakia by using econometric models and external instruments. By using a Proxy-SVAR, we observe that such effects are mostly muted. However, when applying non-linear local projections, we find evidence of asymmetries depending on the state of the economy (expansions vs recessions) and the level of inflation (high vs low inflation). In particular, monetary policy appears to be effective, i.e. contractionary, only in states characterized by recessions and low inflation.

“Short-term forecasting housing investment: An averaging approach from a CEE country”

Conditionaly accepted at the National Bank of Slovakia WPS. Latest version: LINK Media: Global Housing Watch

Abstract: This paper uses a model averaging tool to conditionally predict housing investment in Slovakia up to three quarters ahead. Motivated by the significant modeling uncertainty, I employ a forecast averaging approach that involves two steps. First, it estimates many vector error-correction models (VECM) using a battery of long- and short-run determinants. Second, it applies subset selection based on in-sample and pseudo out-of-sample criteria. This approach produces forecasts of housing investment that widely beat a battery of powerful benchmark models, including VAR, FAVAR, Ridge, and LASSO models. Overall, the results of this study add to the literature showing the good performance of model averaging tools and provide guidance to practitioners modeling and forecasting housing investment in Slovakia and other countries.

“Forecasting business investment: time series vs machine learning models”

Ongoing research. Presentation: LINK

Abstract: The aim of this study is to deploy a tool to conditionally forecast business investment in Slovakia, the sixth most volatile in the euro area. Motivated by significant modeling uncertainty, I propose a model averaging tool that estimates a large and diverse number of VAR, VECM, and machine learning models using a wide set of long- and short-run determinants, as well as soft indicators. After applying subset selection based on in-sample and pseudo-out-of-sample criteria up to three quarters ahead, only a relatively few time series models beat the naïve benchmark AR(1). Alternatively, only seven machine learning (ML) models outperform such benchmark, with an RMSE forecast gain of up to 12 pp., where the top ML model beats the top VECM model by a relatively mild margin, while using much more data.

“The effects of monetary policy shocks in the euro area”

Third PhD chapter. Latest version: LINK

Abstract: This paper estimates the effects of monetary policy and central bank information shocks to subcomponents of GDP and other key macroeconomic variables in the euro area. Additionally, I evaluate whether such effects have changed in the last two decades. To perform such analysis I use an extended version of the SVAR model of Jarocinski and Karadi (2020) and a Proxy-SVAR model. The main findings in this study are as follows. First, purely monetary policy shocks have significantly negative effects on consumption, housing and business investment, having the largest impact on the latter. By contrast, the effects on prices are quite modest, consistent with the literature. Finally, our evidence suggest that the effects of purely monetary policy shocks have changed over time in the euro area. In particular, while during the 2000s the effects are the standard contractionary ones, during the 2010s it seems that the capacity of the ECB to affect economic variables such as business and housing investment and unemployment has been critically weakened.