Time-frequency domain causality of prime building cost and macroeconomic indicators in Ghana: implications for project selection

Abstract

According to the Ghana Statistical Service, construction activity generated US$3.8 billion in 2014, representing about 12.7% of nominal GDP. A key factor that most investors in the construction sector consider in project selection is cost which has escalated over the years. Therefore, we examine the co-movements and causality in the time and frequency domains between the prime building cost index (PBCI) and its disaggregates on one hand and inflation and exchange rate on the other hand in Ghana. The findings revealed that the degree of co-movement between PBCI and inflation was significant at the 5% level and occurred over the long-run, indicating that construction cost drives inflation. The degree of co-movement between the prime build cost index and exchange rate occurred within the short term. These co-movements have policy implications for substituting inflation and exchange rates in project costing by investors in Ghana. The co-movement between the PBCI and inflation in the medium to higher frequencies suggests that inflation risk-averse investors should consider short-term projects. Again, the short to medium frequency co-movements between the PBCI and exchange rate indicates that investors should consider exchange rate risk in long-term projects.

Keywords:

• Prime building cost index
• wavelet analysis
• frequency domain
• time domain
• non-food consumer price index
• exchange rate
• Ghana

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 A phase differences is the difference, measured in radians or degrees, between two series (waves) with the same frequency referenced to the same point in time. Thus, when two series have the same frequency but no phase difference, the series are in phase but if the series have the same frequency and phase differences then they are out of phase.

2 Anti-phase occurs when the phase difference of two series is π radians or 180 degrees.

3 In wavelet application the signals of interest have finite duration. Therefore, as the wavelet approaches the edge of the signal, estimating the convolution needs non-existent data that go beyond the boundary. This creates the edge effects attributed to the incomplete data in the boundary.

4 In case of wavelet coherence: PBCI100 vs CPINF, PBCI100 is the first variable and CPINF is the second variable.