Maurice Larrain

Market efficiency before and after the introduction of electronic trading at the Toronto stock exchange

Abstract The trend toward automation of the trading activity on stock exchanges has spread around the world in recent years. Using rescaled range analysis, we test the effect of automation at the TSE on the markets efficiency using daily and monthly return data. Significant departures from a random walk model were found for selected individual stocks in the daily data. However, monthly returns of various stock indexes were more closely described by a random walk process. Differences in the daily and monthly data may be attributable to the effects of aggregation and indexation on return data. In addition, several simple ‘technical’ trading strategies were compared with a ‘buy and hold’ strategy using daily data for 25 stocks. Despite the presence of non-random patterns in the return data, the technical trading rules could not exploit this information to out-perform the buy and hold strategy. Thus, the presence of deviations from a random walk do not necessarily translate into abnormal performance. Overall, the results suggest automation did not significantly alter the degree of market efficiency at the TSE.

Predictive Ability of the Interest Rate Spread Using Neural Networks

Abstract Interest rates are commonly used as predictors of future economic conditions as measured by industrial production, real gross domestic product and real total business sales (RTBS), as well as through the prediction of recessions in the economy. Recession forecasting is mainly characterized by probit categorical analysis, and there appear to be hardly any neural network research in this area. This paper contributes to the recession forecasting literature using interest rate spreads (the difference between the average yields on 10 year U.S. Treasury bonds and on 3 month U.S. Treasury bills) to forecast the 2007 to 2009 recession with neural network models referenced against regression models. It is shown that neural network models out-performed regression models as evidenced by the R-squared and mean square error performance metrics. Unlike other studies, the change in interest rates is used to compute the interest rate spread. The targeted variable is RTBS. The interest rate spread variable was used to generate three input variables comprising 23, 26, and 29 month leads respectively over RTBS.

Forecasting Purchasing Managers’ Index with Compressed Interest Rates and Past Values

Abstract The purchasing managers’ index (PMI) is a simple subjective survey about the state of the manufacturing sector of the national economy. Its an early indicator of the nations economic strength with effects extending into federal monetary policy and the financial markets. It is a composite index comprising the weighted average of new orders, production, employment, supplier deliveries, and inventories. It has been established that inverted interest rates in 3-month Treasury bills is a predictor of PMI. This study extended the work on the compression of economic and financial predictor variables as well as the relative efficiency of temporal nonlinear neural network models in forecasting economic time series variables. It showed that compressed interest rates and PMI past values are also effective predictors of the future values of PMI. Less than 30% of the wavelet packets coefficients of interest rates were involved in accomplishing the forecasting task. The correlation, root mean square error, normalized root mean square error, mean absolute deviation, and Theil inequality metrics were used to determine the efficacy of the forecasts. The overall PMI forecast produced by the neural network models was relatively better than that produced by the regression models on all metrics except Theil inequality.

Forecasting Real Inventories and the Anomaly of Money Illusion

While the transmission mechanism of inventory behavior in the business cycle has been studied, less effort has been devoted to applied forecasting of inventory change. Inventory fluctuations have accounted for a sizable portion of the changes in U.S. GDP during recessions over the past fifty years. In this paper, we report on out-of-sample forecasts of manufacturing and trade inventories generated by regression and neural network methodology. Our forecasting model is Metzlerian in approach, in that the divergence between actual and targeted sales is hypothesized as the primary cause of inventory imbalance. Our forecasts also rely on the slow adjustment of inventory investment to sales surprises. However, the likely presence of money illusion is a caveat to users, and we address several distortions it introduces to inventory management measures.

The Treasury Bill Rate, the Great Recession, and Neural Networks Estimates of Real Business Sales☆

Abstract This paper analyzes out-of-sample forecasts of real total business sales. We study monthly data from January 1 970 to June 2012. The predictor variable, 3-month Treasury bill interest rate, was used with both the regression (used as a benchmark) and neural network models. The neural network models’, trained in supervised learning with the Levenberg-Marquardt backpropagation through time algorithm, prediction accuracy was confirmed with correlation coefficient and root mean square tests. The activation function used for the focused gamma models of the time-lag recurrent networks in both the hidden and output layers was tanh. The forecast period ranged from January 2006 to June 2012 thus encompassing the past recession. The real business sales variable is one of the indicators used as a coincident index of the U.S. business cycle, and is included among the variables studied by the Federal Reserve to formulate monetary policy. It is thus an important indicator surrogating for real GDP, which is reported quarterly and with a longer time delay. Our analysis shows that recent recessions have increased in duration, so that using a 36-month change to approximate an average cycle in estimating and forecasting is more relevant and accurate than past usage of a 24-month change.

Comparing the Forecasts of Money Demand

Abstract The demand for money depends positively on the price level and real income, and negatively on nominal interest rates and wealth. In addition, since the amount of wealth in an economy is fixed, an individuals or firms wealth is typically tied up between money and bonds. When one of these markets is in equilibrium, so is the other and resultantly money supply is equal to money demand at a particular interest rate. The interest rate affects the movement of the money supply, and Federal Reserve Bank policy influences the short-term interest rate. Monetary policy also affects the price level, while real income in turn affects the movement of money demand. The interaction of money supply and demand leads to a series of equilibriums in the money market. This paper is concerned with the forecasting of money demand changes relative to levels and using price level/inflation, real income, wealth, and interest rate as independent variables. Money demand is approximated by the quantity of M2 money stock, and the price level and interest rate are represented by the consumer price index and 3-month Treasury bills respectively. The forecasting tools used are neural networks and robust multiple linear regression. The efficacy and relative accuracy of the forecasts are determined by performance metrics correlation, root mean square error, and visual analysis. As expected, neural networks yielded a better overall forecast of the changes in money demand.

Housing Starts Forecast of Retail Sales through the 2007-2009 Recession

Abstract The expansion following the 2001 recession was in part stimulated by a boom in housing market investments. Many economists were concerned that a severe drop in residential investments would cause a recession throughout the economy because of residential investments relationship to the gross domestic product and financial markets, and that a decline in housing prices would negatively impact consumer spending. A severe decline in the housing market and the illiquidity in the financial market were both evident by the third quarter of 2007. Some models based on historical data were showing the certainty of a recession in 2007. The best predictors of recessions are respectively interest rate spread, unemployment claims, and building permits. Building permits/new private housing units are a leading economic indicator whereas retail sales are a component of manufacturing and trade sales, which is a coincident economic indicator because it is highly correlated to GDP. This study attempts to use housing starts and past values of retail sales to forecast out-of-sample retail sales values through the period of January 1998 to August 2010, which is inclusive of the recent 2007-2009 recession. If housing Starts are found to be predictive of retail sales, then they are a predictor of a coincident indicator of recessions. This forecast is done in the framework of neural network modeling referenced to robust regression analysis. It was expected that the neural network models would produce better forecasts as ascertained by correlation, root mean square error, and Theil inequality coefficient performance metrics, but the overall result was mixed.

Central bank intervention, the current account, and exchange rates

This paper seeks to explain exchange rate and current account or net foreign assets behavior under central bank foreign exchange rate intervention. To analyze central bank intervention we use the current account-net foreign assets identity, as well as the long-run monetary exchange rate model. The intervention function is one where exchange rate deviations from equilibrium are governed by nonlinear adjustments. That is, exchange rate deviations from their long-run equilibrium are such that the degree of reversion towards equilibrium increases with the size of the deviation from equilibrium. In this type of nonlinear function exchange rates determine the current account, and the current account in turn determines exchange rates. This iterative duality contrasts with several portfolio balance models where exchange rates are a function of trade, but trade is not a function of exchange rates. This two way causality is slightly more complex, but is also analytically richer than assuming that exchange rates change solely in a one step process as targeted by central banks. Managing exchange rates is posited to be an active iterative feedback process where intervention changes the current account, which may in turn make further intervention necessary.

Internal Capital Markets and Patenting in Emerging Growth Firms

Despite the overwhelming theoretical intuition in support of the arguments that internally generated capital should be an important determinant of corporate innovation, very little empirical evidence of this association has been established. In this paper we show that internal capital markets have a positive and significant relationship with patenting of emerging firms. However, for mature firms, the relationship between internal finance and patenting is negative but not significant. Our empirical analysis is grounded on the theoretical modeling of granting a patent as the maturity date of an American real call option, with internal capital and R&D expenses serving to shorten the maturity of the growth option and to speed up innovation.