Skip to main content
Latin American Economic Review

Table 13 Competing models vs. random walk for positive and negative forecast errors

From: Great expectations? evidence from Colombia’s exchange rate survey

 

Positive forecast errors

Negative forecast errors

1-Month

1-Year

1-Month

1-Year

Extrapolative

0.0003 (0.001)

0.19*** (0.035)

−0.0003 (0.001)

0.21*** (0.046)

Adaptive

0.0002 (0.001)

0.17*** (0.033)

−0.0002 (0.001)

0.31*** (0.051)

Regressive

0.0001 (0.001)

0.11*** (0.032)

0.0009 (0.001)

0.06*** (0.022)

Forward discount

0.008*** (0.002)

0.01 (0.015)

−0.001 (0.001)

−0.005 (0.012)

Surveyed expectations

    

All Participants

0.001*** (0.000)

−0.06 (0.087)

0.006*** (0.002)

−0.015 (0.017)

Commercial banks

0.009*** (0.002)

−0.03 (0.094)

0.007*** (0.002)

0.019 (0.021)

Stockbrokers

0.010*** (0.002)

−0.054 (0.082)

0.006*** (0.002)

0.013 (0.015)

Pension funds

0.011*** (0.003)

0.068 (0.139)

0.005*** (0.002)

0.011 (0.025)

  1. Source: authors’ calculations. All estimations correspond to rolling regressions. Values correspond to \(\left( MSPE_{r}-MSPE_{u}\right)\), where \(MSPE_{r}\) and \(MESPE_{u}\) correspond to “restricted” (Random Walk) and “unrestricted” (competing strategies) models. Methodology follows that of Clark and West (2006). Standard errors are in parenthesis
  2. ***, **, * correspond to significance levels of 1, 5 and 10 %, respectively
Back to article page