Multivariate Markov Switching Model Research Articles

Forecasting the instability of Polish banks

The paper presents a formalised procedure of the identification of financial stability EWI (Early Warning Indicators) for the Polish banking sector, in which a two-step procedure was applied. First, the author used a logit model to estimate of the biggest Polish banks’ probabilities of default (PDs). Next, the calculated individual banks’ PDs were used to prepare aggregated domestic banking system stability. In the last step, employing a set of multivariate Markov-switching models with distributed lags (MMSM-DL), the author applied this measure to identify EWI from the candidate macro, private and public debt, banking sector, financial markets and property prices indicators. The best performing EWI were selected with application of area under the receiver operating characteristic (AUROC) metrics and compared with an output of a popular logistic regression (LR) model. To the best author’s knowledge, this article presents for the first time a fully formalised analytical framework based on the MMSM-DL approach that combines microprudential and macroprudential data for the Polish banks financial stability EWI identification. Moreover, the survey supports the hypothesis that the Polish banking sector is stable with use of a formalised econometric procedure.

Disasters, Large Drawdowns, and Long-term Asset Management

Long-term investors are often reluctant to invest in assets or strategies that can suffer from large drawdowns. A major challenge for such investors is to gain access to predictions of large drawdowns in order to precisely design strategies minimizing these drawdowns. In this paper, we describe a multivariate Markov-switching model framework that allows us to predict large drawdowns. We provide evidence that three regimes are necessary to capture the negative trends in expected returns that generate large drawdowns, and we correctly predict conditional drawdowns. In addition, investment strategies based on these models outperform model-free strategies based on the empirical distribution of drawdowns. These results hold within and out of the sample.

A Markov-Switching Model of Inflation in Bolivia

The Bolivian inflation process is analyzed utilizing a time-varying univariate and multivariate Markov-switching model (TMS). With monthly data and, beginning in the late 1930s, inflation is accurately described by a univariate TMS. The intercept for the high-inflation regime is significantly higher than for the low-inflation regime and the actual inflation rate mirrors the smoothing probabilities of the Markov process. Additionally, the predicted duration of each regime closely fits the periods when the country experienced low and inordinate high inflation rates. From a long-run perspective and utilizing a multivariate TMS, the results generally fall in line with what the quantity theory of money predicts. In the high-inflation regime, money growth increases inflation (almost) one-for-one, as classical economics contends. From a short-run perspective and in the high-inflation regime, inflation is almost exclusively explained by a negative output gap. In the low-inflation regime, lagged inflation is the most important determinant of inflation, in line with price stickiness expectations. Partitioning the sources of inflation demonstrate that, from a long-run perspective and in the high inflation regime, differences in inflation are mostly explained by GDP growth; in the low-inflation regime, money growth and velocity growth are the principal factors explaining the variance of inflation. From a short-run perspective, the output gap explains almost all regression variance in the high-inflation regime, and past inflation does the same during times of low inflation, though in both cases the R2 is low which precludes making definite statements about the sources of variability in inflation.

Funding liquidity risk and the low-volatility anomaly: Evidence from the Taiwan stock market

We investigate whether the funding liquidity risk to institutional investors influences the negative relation between expected returns and variance (the ‘‘Low-volatility anomaly’’). With the Taiwan stock market as a setting, we implement a multivariate Markov switching model and use the funding liquidity risk to model the time-varying transition probabilities of the regime-switching process to capture changes in the funding liquidity risk regime. Our evidence documents that the low-volatility anomaly is most pronounced when there is high funding liquidity risk. When there is low funding liquidity risk, however, the low-volatility anomaly has a significant reversal. These results imply that the increased funding liquidity risk due to financial shock transmitted from parent banks is associated with higher selling pressure on institutional investors’ high-volatility stocks, leading to the low-volatility anomaly.

Asymmetric effect of advertising on the Chinese stock market

ABSTRACTIn this paper, we investigate whether investor attention to advertising has an asymmetric effect on Chinese stock returns by using a multivariate Markov switching model with time-varying regime transition probabilities. Using the Chinese stock market as a setting, we obtain lagged conditional volatility from generalized autoregressive conditional heteroskedasticity (GARCH) for modelling the time-varying transition probabilities of the regime-switching process to capture changes in the market regime. Our evidence documents that the high advertising portfolio does earn higher abnormal return than the low advertising portfolio in low-volatility periods. In high-volatility periods, however, the abnormal return is insignificant when the firm increases advertising spending. Our results support the behavioural model argument that in high-volatility period, advertising information diffuses slowly due to cognitive dissonance. Thus, the effect of advertising on stock returns is asymmetric, and it shows statistical significance in low-volatility periods.

THE PROPAGATION OF INDUSTRIAL BUSINESS CYCLES

This paper examines the evolution of the distribution of industry-specific business cycle linkages, which are modeled through a multivariate Markov-switching model and estimated by Gibbs sampling. Using nonparametric density estimation approaches, we find that the number and location of modes in the distribution of industrial dissimilarities change over the business cycle. There is a relatively stable trimodal pattern during expansionary and recessionary phases characterized by highly, moderately, and lowly synchronized industries. However, during phase changes, the density mass spreads from moderately synchronized industries to lowly synchronized industries. This agrees with a sequential transmission of the industrial business cycle dynamics.

The Propagation of Industrial Business Cycles

This paper examines the business cycle linkages that propagate industry-specific business cycle shocks throughout the economy in a way that (sometimes) generates aggregated cycles. The transmission of sectoral business cycles is modelled through a multivariate Markov-switching model, which is estimated by Gibbs sampling. Using nonparametric density estimation approaches, we find that the number and location of modes in the distribution of industrial dissimilarities change over the business cycle. There is a relatively stable trimodal pattern during expansionary and recessionary phases characterized by highly, moderately and lowly synchronized industries. However, during phase changes, the density mass spreads from moderately synchronized industries to lowly synchronized industries. This agrees with a sequential transmission of the industrial business cycle dynamics.

Methods for measuring expectations and uncertainty in Markov-switching models

I develop methods to analyze multivariate Markov-switching models. Formulas for the evolution of first and second moments are derived and then used to characterize expectations, uncertainty, impulse responses, sources of uncertainty, and welfare implications of regime changes in general equilibrium models. The methods can be used to capture the link between uncertainty and the state of the economy. Campbell’s present value decomposition is generalized to allow for parameter instability. Taking into account regime changes is shown to be important for expectations, welfare, and uncertainty. All results are derived analytically and are therefore suitable for structural estimation.

Spillover effect of US monetary policy to ASEAN stock markets: Evidence from Indonesia, Singapore, and Thailand

In this paper, we investigate the spillover effect from US monetary policy to selected ASEAN stock markets by employing Markov-switching models. Based on univariate Markov-switching models, we confirm the existence of two distinct regimes for both US monetary policy and the stock markets. By applying multivariate Markov-switching models, we find that US interest rates have a negative effect on the selected ASEAN stock markets during economic expansion periods. However, this kind of effect disappears during economic crisis periods. Our empirical results indicate that the spillover effect from US monetary policy influences the ASEAN stock markets only during the tranquil period. These results have important implications for the transmission mechanisms of asset price, such as the credit channel, trade channel, and balance sheet channel.

Global imbalances, cross-market linkages, and the financial crisis: A multivariate Markov-switching analysis

Based on multivariate Markov-switching models, this paper presents new results on the interactions between global imbalances, credit spreads, housing markets, macroeconomic variables, commodities and equities during Q1-1987/Q1-2011. We show that rising global imbalances and the uncontrolled development of the US mortgage and housing markets have been deeply destabilizing the economy, with various shocks impacting subsequently equity markets and macroeconomic variables. But we also uncover, surprisingly, that the cross-market linkages with the commodity markets are strong. Finally, we identify that the US housing market lies at the epicenter of the crisis through its multiple and highly significant interactions with the other variables in the system (including the global imbalances). Sub-samples and alternative time series estimates are provided to check the statistical congruency of the various models.

When does investor sentiment predict stock returns?

We examine the asymmetry in the predictive power of investor sentiment in the cross-section of stock returns across economic expansion and recession states. We test the implication of behavioral theories and evidence that the return predictability of sentiment should be most pronounced in an expansion state when investors' optimism increases. We segregate economic states according to the NBER business cycles and further implement a multivariate Markov-switching model to capture the unobservable dynamics of the changes in the economic regime. The evidence suggests that only in the expansion state does sentiment perform both in-sample and out-of-sample predictive power for the returns of portfolio formed on size, book-to-market equity ratio, dividend yield, earnings-to-price ratio, age, return volatility, asset tangibility, growth opportunities, and 11 widely documented anomalies. In a recession state, however, the predictive power of sentiment is generally insignificant.

REGIONAL BUSINESS CYCLE PHASES IN TAIWAN

AbstractThis paper uses both univariate and multivariate Markov‐switching models to explore the characteristics of the regional business cycle phases in Taiwan. In particular, this paper provides novel evidence regarding the “turning‐point identification” puzzle in Taiwan that to date has not been fully resolved. We find that the regime‐switching behaviour of the common regional business cycle is consistent (or even precedes recession) with the business cycle expansion and contraction indicators used by the Council for Economic Planning and Development. This finding suggests that the common regional business cycle identifies the turning point after the 1990s. This finding also suggests that the employment growth index would be a great indicator for monitoring the business cycle in Taiwan.

When Does Investor Sentiment Predict Stock Returns?

We examine the predictive effect of sentiment on the cross-section of stock returns across different economic states. The degree of mispricing and the subsequent price correction can be different between economic expansion and recession because of the limits of arbitrage and short sale constraints. The predictive ability of sentiment is asymmetric between different states of the economy. We implement a multivariate Markov-switching model to characterize the economic states. Conditional on the identified economic states, we use the lagged sentiment proxy to forecast the portfolio returns related to small stocks, non-earning stocks, growth stocks, and non-dividend-paying stocks. We find that only in the expansion state does sentiment performs both in-sample and out-of-sample predictive power on these categories of stocks. When an expansion state has high sentiment, these categories of stocks earn relatively low subsequent returns. The predictive ability of sentiment can not be attributed to time-variation in the market beta driven by investor sentiment.

Can VAR Models Capture Regime Shifts in Asset Returns? A Long-Horizon Strategic Asset Allocation Perspective

We examine whether simple VARs can produce empirical portfolio rules similar to those obtained under a range of multivariate Markov switching models, by studying the effects of expanding both the order of the VAR and the number/selection of predictor variables included. In a typical stock bond strategic asset allocation problem on US data, we compute the out-of-sample certainty equivalent returns for a wide range of VARs and compare these measures of performance with those typical of non linear models that account for bull-bear dynamics and characterize the differences in the implied hedging demands for a long-horizon investor with constant relative risk aversion preferences. In a horse race in which models are not considered in their individuality but instead as an overall class, we find that a power utility investor with a constant coefficient of relative risk aversion of 5 and a 5-year horizon, would be ready to pay as much as 8.1% in real terms to be allowed to select models from the MS class, while analogous calculation for the whole class of expanding window VAR leads to a disappointing 0.3% per annum. We conclude that most (if not all) VARs cannot produce portfolio rules, hedging demands, or out-of-sample performances that approximate those obtained from equally simple non-linear frameworks.

Markov-switching and the Beveridge–Nelson decomposition: Has US output persistence changed since 1984?

Abstract We show that, for a class of univariate and multivariate Markov-switching models, exact calculation of the Beveridge–Nelson (BN) trend/cycle components is possible. The key to exact BN trend/cycle decomposition is to recognize that the latent first-order Markov-switching process in the model has an AR(1) representation, and that the model can be cast into a state-space form. Given the state-space representation, we show that impulse-response function analysis can be processed with respect to either an asymmetric discrete shock or to a symmetric continuous shock. The method presented is applied to Kim, Morley, Piger’s [Kim, C.-J., Morley, J., Piger, J., 2005. Nonlinearity and the permanent effects of recessions. Journal of Applied Econometrics 20, 291–309] univariate Markov-switching model of real GDP with a post-recession ‘bounce-back’ effect and Cochrane’s [Cochrane, J.H., 1994. Permanent and transitory components of GNP and stock prices. Quarterly Journal of Economics 109, 241–263] vector error correction model of real GDP and real consumption extended to incorporate Markov-switching. The parameter estimates, the BN trend/cycle components, and the impulse-response function analysis for each of these empirical models suggest that the persistence of US real GDP has increased since the mid-1980’s.

Multivariate Markov Switching with Weighted Regime Determination: Giving France More Weight than Finland

This article deals with using panel data to infer regime changes that are common to all of the cross section. The methods presented here apply to Markov switching vector autoregressions, dynamic factor models with Markov switching and other multivariate Markov switching models. The key feature we seek to add to these models is to permit cross-sectional units to have different weights in the calculation of regime probabilities. We apply our approach to estimating a business cycle chronology for the 50 U.S. States and the Euro area, and we compare results between country-specific weights and the usual case of equal weights. The model with weighted regime determination suggests that Europe experienced a recession in 2002-03, whereas the usual model with equal weights does not.

What Tames the Celtic Tiger? Portfolio Implications from a Multivariate Markov Switching Model

We use multivariate regime switching vector autoregressive models to characterize the time-varying linkages among the Irish stock market, one of the top world performers of the 1990s, and the US and UK stock markets. We find that two regimes, characterized as bear and bull states, are required to characterize the dynamics of excess equity returns both at the univariate and multivariate level. This implies that the regimes driving the small open economy stock market are largely synchronous with those typical of the major markets. However, despite the existence of a persistent bull state in which the correlations among Irish and UK and US excess returns are low, we find that state comovements involving the three markets are so relevant to reduce the optimal mean-variance weight carried by ISEQ stocks to at most one-quarter of the overall equity portfolio. We compute time-varying Sharpe ratios and recursive mean-variance portfolio weights and document that a regime switching framework produces out-of-sample portfolio performance that outperforms simpler models that ignore regimes. These results appear robust to endogenizing the effects of dynamics in spot exchange rates on excess stock returns.

A useful tool for forecasting the Euro-area business cycle phases

Based on a novel extension of existing multivariate Markov-switching models, we provide the reader with a useful tool for analyzing current business conditions and making predictions about the future state of the Euro-area economy in real time. Apart from the Industrial Production Index, we find that the European Commission Industrial Confidence Indicator, which is issued with no delay, is very useful for constructing the real-time predictions.

Forecasting VaR and Expected Shortfall using Dynamical Systems: A Risk Management Strategy

Using non-parametric and parametric models, we show that the bivariate distribution of an Asian portfolio is not stable along all the period under study. We suggest several dynamic models to compute two market risk measures, the Value at Risk and the Expected Shortfall: the RiskMetrics methodology, the Multivariate GARCH models, the Multivariate Markov-Switching models, the empirical histogram and the dynamic copulas. We discuss the choice of the best method with respect to the policy management of bank supervisors. The copula approach seems to be a good compromise between all these models. It permits taking financial crises into account and obtaining a low capital requirement during the most important crises.

What tames the Celtic Tiger? Portfolio implications from a Multivariate Markov Switching model

We use multivariate regime switching vector autoregressive models to characterize the time-varying linkages among the Irish stock market, one of the top world performers of the 1990s, and the US and UK stock markets. We find that two regimes, characterized as bear and bull states, are required to characterize the dynamics of excess equity returns both at the univariate and multivariate level. This implies that the regimes driving the small open economy stock market are largely synchronous with those typical of the major markets. However, despite the existence of a persistent bull state in which the correlations among Irish and UK and US excess returns are low, we find that state comovements involving the three markets are so relevant to reduce the optimal mean–variance weight carried by ISEQ stocks to at most one-quarter of the overall equity portfolio. We compute time-varying Sharpe ratios and recursive mean–variance portfolio weights and document that a regime switching framework produces out-of-sample ...