Recession prediction for US border economies. Can probit models predict recessions at an MSA level including bi-national variables?

Economic recession prediction has received substantial attention in recent years. The topic is important. Unemployment, business bankruptcies, and personal bankruptcies all increase during business cycles downturns. Identifying appropriate indicators for recessions is a difficult task. Other variables have been used in prior studies. These variables have included the term structure of the yield curve (Fama 1984; Mishkin 1988a) and the difference between the yields of long-run treasury bonds and short run treasury bills (yield curve) (Estrella and Mishkin 1996; Dueker 1997; Kauppi and Saikkonen 2008; Nyberg 2010). Estrella and Mishkin (1998) compare the effectiveness of different economic variables that are supposed to have potential predictive information. These variables are interest rates, interest rate spreads, stock price indexes, monetary aggregates and the yield curve. The yield curve has proved to be the indicator with the most predictive power. The major caveat of the yield curve is that it predicts recessions better at longer lead times (more than 6 months) than most other financial variables. However, if a recession is expected in the short-run (6 months or less), the predictive performance of the yield curve worsens. In contrast, the other indicators often prove to be more reliable in the short-run. The yield curve has also proven to have good regional predictive power (Gauger and Schunk, 2002; Shoesmith, 2003). In these models, the yield spread is used as the independent variable with a dummy variable for recession as the dependent variable. The results show that for regional economies, the yield curve also helps forecast recessions. This paper examines the ability of the yield curve (difference between the 10-year Treasury bond rate and the 3-month Treasury bill rate) to predict recessions for the largest American economies along the US - Mexico border. Because of the economic interrelation between these border economies, other variables will also be included in the sample. Other variables include the peso – dollar exchange rate and the yield curve for Mexico (difference between the 1-year Treasury bond rate and the 1-month Treasury bill rate)