This paper proposes a new regression model — a smooth transition mixed data sampling (STMIDAS) approach — that captures recurrent changes in the ability of a high frequency variable in predicting a variable only available at lower frequency. The model is applied to the use of …nancial variables, such as the slope of the yield curve, the short-rate and stock returns, to forecast US output growth both in- and out-of-sample. I …nd evidence that the use of the predictor sampled weekly improves output growth forecasts, which may also be improved when changes in …nancial variables’predictive power are considered.

In 2005 all ECB publications will feature a motif taken from the €50 banknote.

In this paper we propose a method for determining the number of regimes in threshold autoregressive models using smooth transition autoregression as a tool. As the smooth transition model is just an approximation to the threshold autoregressive one, no asymptotic properties are claimed for the proposed method. Tests available for testing the adequacy of a smooth transition autoregressive model are applied sequentially to determine the number of regimes. A simulation study is performed in order to find out the finite-sample properties of the procedure and to compare it with two other procedures available in the literature. We find that our method works reasonably well for both single and multiple threshold models. Key words: Model specification, model selection criterion, nonlinear modelling, sequential testing, switching regression.

A major question in the literature on the classical gold standard concerns the efficiency of international arbitrage. Most authors have examined efficiency by looking at the spread of the gold points, gold-point violations, the flow of gold in profitable or unprofitable directions, or by tests of various asset market criteria, including speculative efficiency and interest arbitrage. These studies have suffered from many limitations, both methodological and empirical. We offer a new methodology for measuring market integration based on nonlinear theoretical models applied using the techniques of threshold autoregressions. We improve the empirical basis for investigation by compiling a new, high-frequency series of continuous daily data from 1879 to 1913. Using data at this frequency we can derive reasonable econometric estimates of the implied gold points and price dynamics. The changes in these measures over time provides an insight into the evolution of market integration.

In this paper we propose a testing procedure for assessing the presence of threshold effects in nonstationary Vector autoregressive models with or without cointegration. Our approach involves first testing whether the long run impact matrix characterising the VECM type rep-resentation of the VAR switches according to the magnitude of some threshold variable and is valid regardless of whether the system is purely I(1), I(1) with cointegration or stationary. Once the potential presence of threshold effects is established we subsequently evaluate the cointegrating properties of the system in each regime through a model selection based approach whose asymptotic and finite sample properties are also established. This subsequently allows us to introduce a novel non-linear permanent and transitory decomposition of the vector process of interest. ∗ We wish to thank the Spanish Ministry of Education for supporting this research under grants No. SEC01-0890

A generalization of the self-exciting threshold autoregressive model where seasonality is modelled to be deterministic and regime-dependent is formulated, and inference in this seasonal SETAR (SeaSETAR) framework is discussed. The model is tted to quarterly, seasonally unadjusted unemployment rate data from US and Japan. Such a specication is signicant when tested against a linear alternative for both of the datasets considered, and the null of equal seasonal eects across regimes is strongly rejected for both countries. As a further exercise on the convenience of regime-dependent seasonality modelling, out of sample forecasts up to one year (four steps-ahead) are generated from the SeaSETAR models using Monte Carlo and bootstrapping methods. When compared to the out of sample performance of a competing linear model, the nonlinear specication is superior for all forecasts horizons in the case of Japan unemployment rate and for three and four steps-ahead in the case of US unemployment rate. Keywords: Unemployment, Nonlinear Time Series Models; SETAR Models; Seasonality, Forecasting. JEL Classication: C53, C52, C22 1 1

this paper we use these shooting records to address the following questions: 769 grouse cycles 2004 British . What proportion of grouse time-series are asymmetrical and/or non-linear? . How does asymmetry manifest itself within the time-series? . Is there geographical variation in patterns of asymmetry and non-linearity? . Are variations in patterns of asymmetry and nonlinearity related to variations in grouse population density, growth and decline rates or with anticipated patterns in intensity of parasitism as predicted by rainfall? Methods The time-series used came from moors located in all the major regions of grouse habitats in the United Kingdom where shooting has occurred for at least 32 years. Two hundred and eighty-nine grouse time-series were included, de-trended and patched or split, as described in Haydon et al. (2002). We excluded from subsequent analyses those time-series that were not distinguishable (at the 5% level) either from white noise using the Ljung--Box test, or from a zero-th order autoregressive process (AR0, see below). For nine populations there were multiple time-series of which nonewere either white noise or AR0, and in these cases the longest time-series was selected for analysis. For the purposes of model fitting and non-linear analysis, time-series were normalized using Box--Cox transformations by taking the value of the exponent that was the maximum likelihood for each timeseries (see Results) as described by Sokal &Rohlf (1981) and then standardized to zero mean and unit variance. Time reversibility We applied Rothman's (1992) TR test to time-series to determine time-reversibility, which examines the equality of the bi-covariances for various values of lag, k (where x t is the transformed time-series, an...

Daily financial returns (and daily stock returns, in particular) are commonly modeled as GARCH(1,1) processes. Here we test this specification using new model evaluation technology developed in Ashley and Patterson (2006), which examines the ability of the estimated model to reproduce features of particular interest: various aspects of nonlinear serial dependence, in the present instance. Using daily returns to the CRSP equally weighted stock index, we find that the GARCH(1,1) specification cannot be rejected; thus, this model appears to be reasonably adequate in terms of reproducing the kinds of nonlinear serial dependence addressed by the battery of nonlinearity tests used here.

Actively managed mutual funds, in general, underperform a passive benchmark; however, some recent studies find they, in fact, outperform the benchmark in bad economic states. I examine whether a state dependent risk shifting behavior of mutual fund managers contributes to this performance difference across states, and find supportive evidence. As shown in prior studies, the risk shifting behavior is motivated by a non-linear flow-performance relationship. Using a piece-wise linear regression, I demonstrate that the non-linearity exists mainly in good states; whereas in bad states, the flow-performance relationship is close to linear. Thus, non-zero risk shifting incentives are only expected in good states. I empirically measure these incentives in good states, and show that managers do react to the “gambling ” (i.e., positive) incentives. In addition, higher “gambling” Actively managed mutual funds, in general, underperform passive benchmarks, net of fees (e.g., Jensen (1968), Malkiel (1995), and Fama and French (2010)). This underperformance is typically interpreted as evidence that mutual fund managers as a group lack the ability to identify mispriced

Threshold models have gained much recent attention in applied economics for modeling nonlinear behavior. The appeal for these models is in part due to the observable pattern that many economic variables follow, such as asymmetric adjustment towards equilibrium. Recent developments in model specification derive error-correction models as a specific type of threshold models. This paper summarizes the developments in threshold modeling over the past two decades and reviews a sample of empirical works in agricultural economics. Guidance is provided for obtaining software programs.