Non-nested Tests Research Articles

The empirical relationship between aggregate consumption and security prices in Australia

This paper provides an empirical examination of the consumption-based capital asset pricing model (CCAPM) using Australian data. Multivariate tests of the CCAPM using actual quarterly consumption data indicate good support for the CCAPM. In a second phase of tests, the maximum correlation portfolio (MCP) is used to proxy the consumption variable using monthly data. Generally, the results of these tests are very mixed. Finally, for non-nested tests of the CCAPM versus the CAPM the evidence favours the traditional CAPM using quarterly data, but is inconclusive using monthly data.

Price determination and rational expectations

A forward looking specification of the error correction model is estimated for Icelandic inflation for the period 1962–1993. Our results are consistent with the view that prices are set by rational, forward looking firms, reacting to expected future innovations in production costs. The forward model is statistically well determined, and the rational expectations, cross-equation restrictions are accepted. We find that the model has stable and plausible parameter values with a stable VAR expectations generating process. Non-nested tests indicate that the forward looking specification cannot be rejected against a statistically well determined backward looking specification. © 1998 John Wiley & Sons, Ltd.

Motivation Models of Australia's Bilateral Aid Program: The Case of Indonesia*

This paper is concerned with the question: what motivates Australia's provision of overseas aid to Indonesia? The econometric results on time series data obtained by testing a recipient need and a donor interest model suggest both these concerns are relevant to Australian aid decisons on Indonesia To determine which model dominates, non-nested tests were conducted; these indicated acceptance of the recipient need model and rejection of the donor interest model. Thus Australia's aid to Indonesia was found to be determined by recipient need rather than donor interest. * The authors wish to thank an anonymous referee for suggestions.

A non-nested test of GARCH vs. EGARCH models

This study uses a Cox-type non-nested test. The test is obtained using Monte Carlo hypothesis tests with the log likelihood ratio as the test statistic. Monte Carlo methods are used to obtain the probability of a larger value of the test statistic under the null hypothesis. The approach used does not rely upon asymptotic normality. Using the maximum likelihood estimation technique, two competing time series models, generalized autoregressive conditional heteroscedasticity (GARCH) and exponential GARCH (EGARCH) models of daily spot prices of Deutsche mark are estimated. Using Monte Carlo hypothesis tests, then, p-values for GARCH vs. EGARCH models are calculated. The EGARCH model cannot be rejected, while the GARCH model is rejected.

Generalized Models of Japanese Demand for Fish

AbstractGiven a relative lack of knowledge about Japanese consumer preferences for fish, Japanese fish demand is modeled using both Marshallian (ordinary) and inverse demand systems, each of which nests a number of competing specifications. Results indicate that the inverse demand systems dominate the ordinary demand systems in forecasting performance and in nonnested tests. The inverse system suggests that Japanese fish prices are less responsive to changes in consumption than found in previous studies.

The market efficiency hypothesis: The case of coffee and cocoa futures

This study tests the market efficiency hypothesis for coffee and cocoa futures using daily data for contracts with a maturity of 2 and 6 months. The hypothesis is tested sequentially. The first condition is that future spot and futures prices be cointegrated. If this condition is maintained, market efficiency requires the cointegrating vector to support a (0, 1) restriction that can be likened to an unbiasedness condition. Finally, market efficiency imposes zero restrictions on the parameters of the variables expressed in first differences in the specification of the error-correction representation of the relationship between future spot and futures prices. Brenner and Kroner [Brenner, R., Kroner, K., 1995. Arbitrage, cointegration, and testing the unbiasedness hypothesis in financial markets. Journal of Financial and Quantitative Analysis 30, pp. 23–42] argue that the cointegration condition is rarely met in practice. They attribute this outcome to potentially non-stationary net cost-of-carry which would make the parameters of the cointegration relation unstable. It is for this reason that Hansen's tests [Hansen, B.E., 1992. Tests of parameter instability in regressions with I(1) processes. Journal of Business and Economic Statistics 10] about the stability of the parameters in cointegration regressions were used to supplement more traditional cointegration tests. Johansen and Juselius' cointegration tests [Johansen, S., Juselius, K., 1993. Testing structural hypotheses in a multivariate cointegration analysis of the PPP and the UIP for UK, Journal of Econometrics 53] could not reject cointegration for all four contracts while Hansen's LC test favored cointegration only for the cocoa contracts. Nested and non-nested testing procedures were used to test the (0, 1) restriction on the cointegration vector. Unbiasedness was found to be robust across testing procedures. However, further testing about the specification of the error-correction representation revealed the existence of important short run deviations from unbiasedness. Even though these results hold strictly for a rather limited number of contracts and commodities, they are encouraging for futures markets advocates in developing countries.

Taxation and Corporate Investment in Physical and Knowledge Capital: Tests of Perfect Competition Versus Market Power

This article specifies an intertemporal model of firm behavior consistent with rational expectations, assuming either perfect competition or market power. Non-nested hypothesis tests were used to test the two versions of this model against each other. The test results do not support perfectly competitive industries and instead suggest that firms in these industries exerted some market power. The effects of tax allowances for physical and research and development (R&D) capital on the firms' decisions were computed. From the public sector's point of view, neither of these tax incentives was cost-effective for Turkey's chemical industries during the sample period. Their costs to the treasury in terms of foregone revenues were higher than the investment stimulated by these measures. For Turkey's petroleum derivatives industry, the benefits of the tax allowance for physical investment (increased physical investment) exceeded its costs (foregone tax revenue). The tax allowance for R&D investment, however, was not cost-effective for the petroleum derivatives industry either.

Capital depreciation and investment demand

Little is known about the way in which capital depreciates. Researchers from Gnliches (1960) to Hulten, Robertson, and Wykoff(1989) have inferred patterns of depreciation from the age-price profile of used capital goods. An alternative method, pioneered by Coen (1975), infers depreciation patterns from the behavior of investment expenditures. This study exploits Coen's investment approach. Numerous capital stocks, generated for alternative combinations of depreciation methods and service lives, are fitted to a standard neoclassical model of investment demand. Several newly developed non-nested tests are used to determine which pattern of depreciation is best. This study provides, for the first time, findings relating to whether the differences in the ability of alternative patterns of depreciation to explain actual investment expenditures are statistically significant.Aggregate data for U.S. manufacturing industry's investment in producers' durable equipment and nonresidential structures for the period 1949–1982 are analysed. The results indicate that the hypothesis that all depreciation is geometric may be rejected, for both structures and equipment. However, it is not possible to accept any single pattern as representing all depreciation.

ARCH and Bilinearity as Competing Models for Nonlinear Dependence

In this article we consider whether the wide acceptance of autoregressive conditional heteroscedasticity (ARCH) models may be at the expense of other nonlinear processes, such as bilinear models. We first propose a joint test for ARCH and bilinearity. A nonnested test is then suggested to determine whether nonlinear dependence should be attributed to ARCH or bilinearity. The tests are then applied to three series. When generalized ARCH (GARCH) models are taken as the null hypothesis, we fail to reject it for all the data series. When bilinearity is taken as the null, however, it is rejected in two cases. Moreover, an out-of-sample forecasting exercise shows that the GARCH model is superior. The results, therefore, indicate a strong preference for the GARCH model.

Testing nested and non-nested periodically integrated autoregressive models

Periodically integrated time series require a periodic differencing filter to remove the stochastic trend. A non-periodic integrated time series needs the first-difference filter for similar reasons. When the changing sea- sonal fluctuations for the non-periodic integrated series can be described by seasonal dummy variables for which the corresponding parameters are not constant within the sampie, such a series may not be easily & stinguished from a periodically integrated time series. In this paper, nested and non-nested testing procedures are proposed to distinguish between these two alternative stochastic and non-stochastic seasonal processes, When it is assumed there is a single unknown structural break in the seasonal dummy parameters. Several empirical examples using quarterly real macroeconomic time series for the United Kingdom illustrate the nested and non-nested approaches.

Microeconometric models of infrequently purchased goods: An application to household pork consumption

The double-hurdle and infrequency-of-purchase models are generalized with the inverse hyperbolic sine transformation in the dependent variable. The resulting specifications feature more flexible parameterization and error distributions than the untransformed models. Using the 1987–88 Nationwide Food Consumption Survey data on household pork consumption, a nonnested test suggests that the IHS double-hurdle model provides better characterization of the data-generating process than the IHS infrequency-of-purchase model but the elasticities derived from these models are similar. Own-price effects on the probability and level of consumption are negative and significant, but the elasticities are small. Income and cross-price effects are not significant. Household age composition, education, gender of meal planner, and race are among the demographic variables that affect consumption.

Some Tests of Alternative Accessibility Measures: A Population Density Approach

This paper presents nine alternative accessibility measures and evaluates their usefulness in explaining population distribution. It first evaluates these measures by using the criterion of maximum explanatory power in standard regression analysis. It then performs non-nested tests on these nine measures. The paper concludes that gravity-type accessibility measures generally perform better than other measures. The most commonly used measure, size of opportunity weighted by the reciprocal of distance, is not statistically surpassed by any other measures. Cumulative opportunity within the average commuting distance and the distance to the central business district are the poorest accessibility measures.

Relative Prices in the UK Personal Sector Money Demand Function

Nonparametric demand analysis is used in order to establish admissible monetary aggregates and also to establish a potential role for relative prices in the money demand function. A conventional money demand specification is then contrasted with a money demand function which contains the relative prices of durables, nondurables, and services as additional regressors. Evidence from long-run cointegration analysis, short-run error correction models, and nonnested testing all confirm a role for relative prices in the U.K. personal sector's money demand function. Copyright 1996 by Royal Economic Society.

Testing between alternative wage-employment bargaining models using Belgian aggregate data

In this paper, we discriminate among alternative models of bargaining for wages and employment (right-to-manage, efficient bargaining and general bargaining models) using Belgian aggregate data. We estimate the ECM representation of a dynamic employment equation for each model using Engle-Granger's two-step estimation procedure. We use Phillips-Hansen's FME to obtain long-run parameter estimators which are optimal and asymptotically normally distributed. On the basis of non-nested tests, both the right-to-manage and the hypothesis that wage-employment negotiations are efficient are rejected in favour of the general bargaining model where outcomes are inefficient.

A non-nested test of the AIDS vs. the translog demand system

A non-nested test is used to examine the relative explanatory power of the AIDS and translog demand system. Empirical results based on Chinese urban household data suggest that these two demand systems are not significantly different in explanatory power and elasticity estimates.

The Carlson-Parkin method applied to NZ price expectations using QSBO survey data

The standard Carlson-Parkin ( Economica, 1975, 42, 123–138) method is used to estimate the parameters of producers' prices expectations. Non-nested tests suggest that the Carlson-Parkin method is superior to the use of the simple balance statistic. A non-linear extension of the Carlson-Parkin technique is estimated, and no evidence is found to support the hypothesis that producers' price expectations are drawn from skewed or kurtotic distributions.

Model choice and structural specification for Canadian meat consumption

Choice of functional form and structural change specification may each influence demand parameter estimates. Joint non-nested testing of both is applied for the linearised almost ideal and Rotterdam models with and without structural change, incorporated using a gradual transition specification. The test of model choice without structural change is inconclusive; the test of models with structural change shows that, for Canadian meat consumption, the gradual-transition almost ideal model is preferred over the gradual-transition Rotterdam model. The effect of functional form on demand elasticities for meats is relatively minor, but the structural change provision substantially affects these estimates. Copyright 1996 by Oxford University Press.

Measuring angling quality in count data models of recreational fishing

This paper develops a methodology for investigating different specifications of angling quality in recreational demand models. The methodology is used to compare three alternative specifications of the relationship between angling demand and site quality. The site quality alternatives include chemical concentrations, biological stress indices based on the chemicals, and catch per unit effort. The three alternatives are differentiated using a likelihood ratio non-nested testing procedure.

The choice of functional form and estimation of banking inefficiency

The effects of alternative functional forms on parametric estimation of a multi-product cost function are examined using a sample of rural US banks. Functional forms considered for possible choice are three versions of second-order Taylor-series expansions (or locally-flexible functional forms–translog, normalized quadratic, and generalized Leontief. The generalized quadratic Box–Cox transformation and the likelihood dominance criterion are used as vehicles for conducting nested and non-nested tests. Although all three forms are rejected by the nested test, the likelihood dominance criterion indicates that the translog is clearly preferred over the other two as the functional form of choice. Using either of the other functional forms would substantially alter conclusions about both scale elasticities and inefficiencies, including the impact of bank size on these measures.

Consistent Specification Testing Via Nonparametric Series Regression

This paper proposes two consistent one-sided specification tests for parametric regression models, one based on the sample covariance between the residual from the parametric model and the discrepancy between the parametric and nonparametric fitted values ; the other based on the difference in sums of squared residuals between the parametric and nonparametric models. We estimate the nonparametric model by series regression. The new test statistics converge in distribution to a unit normal under correct specification and grow to infinity faster than the parametric rate (n -1/2 ) under misspecification, while avoiding weighting, sample splitting, and non-nested testing procedures used elsewhere in the literature. Asymptotically, our tests can be viewed as a test of the joint hypothesis that the true parameters of a series regression model are zero, where the dependent variable is the residual from the parametric model, and the series terms are functions of the explanatory variables, chosen so as to support nonparametric estimation of a conditional expectation. We specifically consider Fourier series and regression splines, and present a Monte Carlo study of the finite sample performance of the new tests in comparison to consistent tests of Bierens (1990), Eubank and Spiegelman (1990), Jayasuriya (1990), Wooldridge (1992), and Yatchew (1992) ; the results show the new tests have good power, performing quite well in some situations. We suggest a joint Bonferroni procedure that combines a new test with those of Bierens and Wooldridge to capture the best features of the three approaches.