Printer Friendly

Testing the Philips curve hypothesis for Nigeria: are there likely implications for economic growth?

1. Introduction

Macroeconomic response to monetary policy remains critical to the debate on the Philips curve hypothesis and its backward and forward looking variants. Roeger and Herz (2012) pointed out the difference between the backward and forward-looking Phillips curve concerning and the macroeconomic effects of monetary policy shocks. On the backward looking Phillips curve, they noted that it predicts that a positive short-run response of the economic or output is followed by a period in which output is below baseline and that, the cumulative output effect would be zero. On the other hand, the forward-looking variant would imply that there is a positive cumulative output effect from monetary policy shock. Though empirical evidence from Roeger and Herz (2012) showed that the output effects of money are consistent with the forward-looking variant, studies such as Rudd and Whelan (2006) had questioned the relevance of the forward looking term in the hybrid specification hence, debunking backward looking view. In Roeger and Herz (2012) however, it was noted that the macroeconomic response to monetary policy changes remains a vital aspect to be examined.

The need to reduce unemployment and achieve low/single digit inflation rate, have remained the focus of various governments in Nigeria since the return to democracy in 1999. This need has also been expressed through various economic reforms and blue prints such as the National Economic Empowerment and Development Strategy (NEEDs I and II), Nigeria's Transformation Agenda and national vision 20:2020 blue print (NPC, 2004, and NPC, 2007). Curtailing the response to monetary adjustment has also been the basis upon which several monetary policy committee decisions have been based in altering the nation's monetary policy rate in recent times. The aim of this paper therefore, is to validate the Phillips Curve Hypothesis in Nigeria and to examine the implication for Nigeria's economic growth.

The paper is divided into five sections. Section one discussed the problem statement and objectives of the paper. Section two, reviews empirical literature and theoretical literature related to the study, while section three discusses the research methodology. In section four, the results from the analysis of data is discussed, and lastly, the summary of findings as well as their implication for policy is presented in section five.

2. Literature Review

Empirical studies on the Philips curve hypothesis vary in their conclusions. Some researchers found the significant trade-off relationship between unemployment rate and inflation rates, others do not. Alogoskoufis and Smith (1991) showed evidence to support the "Lucas critique" which denied the existence of trade-off relationship. By contrast, King and Watson (1994) tested the existence of the Phillips curve for the U.S. Their findings provided empirical support to the existence of the trade-off relationship between unemployment and inflation in the USA over the researched period. Hansen and Pancs (2001) examined the existence of the Phillips curve in Lativa. They also found that there was a negative correlation between the unemployment rate and the actual inflation rates. Bhanthumnavin (2002) tested the Phillips curve hypothesis for Thailand. The results suggested that the Phillips curve relationship emerged in Thailand after the onset of the Asian crisis. The emergence of the Philips curve hypothesis after the crisis supports the argument that new data sets and estimation techniques could validate the Philips curve hypothesis. This could also be said of the study by Islam et al. (2003), who examined the Philips curve hypothesis for the US from 1950 to 1999, and found that there was a weak long-run (cointegrating) relationship and long-run causality between unemployment and inflation. According to Gokal and Hanif (2004), one of the fundamental objectives of macroeconomic policies in industrialized and less developed economies is to sustain or achieve high economic growth while keeping inflation low. They however, noted that the relationship between inflation and economic growth for specific countries would be critical for achieving the desired outcome. They tested the inflation-economic growth relationship for Fiji (a developing country) and found that there was a weak and negative correlation between inflation and growth. Their study suggested that, low inflation was not sufficient to bring about sustainable economic growth.

Ogbokor (2005) tested the short run Philips curve hypothesis for Namibia, using data from 1991 to 2005. Linear and logarithmic regression models were the estimation technique employed for the analysis. The results offered support for the presence of stagflation in Namibia: a case of inflation and unemployment moving in the same direction. Contrary to the underlying philosophy of the Phillips curve that inflation and unemployment are inversely related. Moosa (2008) tested Okun's law in selected Arab countries by estimating the relationship between economic growth and unemployment. It was found was that unemployment does not determine economic growth. Their comparative evidence however found that unemployment was a significant determinant of economic growth in the US, Europe and Japan. They gave three possible reasons why the Okun's law was not valid for the selected Arab states: (1) that unemployment in these countries was not cyclical, but rather structural and/or frictional. Structural unemployment results from changes in the economy that are not matched by changes in education and training, (2) rigidity of the labor markets in these countries, dominated by the government as the prime source of demand for labor, and (3) the structures of the economies, which is dominated by the government and perhaps one sector such as oil (Moosa, 2008: 9-10). The study however suggested that, the structure of the labor market would influence the outcome of the relationship between unemployment and economic growth.

Garcia (2010) examined the relevance of the Philips curve hypothesis for monetary policy in Nigeria by employing Bayesian econometrics and impulse response functions. The study found that the Nigerian Central Bank could control the inflation rate through a Philips curve framework and concluded that a full-fledge inflation targeting regime in Nigeria could be feasible. The study however was silent on the inflation-unemployment relationship and the conclusion that the Philips curve hypothesis was useful for monetary policy in Nigeria implied that, the Philips curve hypothesis had monetary policy as its end. While inflation targeting would be a useful pass through for the central bank's monetary policy, it would be important to examine the implication of the inflation and unemployment on economic growth as well.

The theoretical background for the Philips curve hypothesis is premised on the wage curve model, the mark-up pricing rule and adaptive expectation model, and Okuns law (Harris and Sylverstone, 2001:1). Economic theory suggests that certain variables have long-run equilibrium relationship. Although the variables may drift away from equilibrium for a while, economic forces or government actions may be expected to restore equilibrium (Keele and DeBoef, 2004). While there are several techniques to test for cointegration, econometricians align towards the model that describes both shortrun dynamics and the long-run equilibrium simultaneously. Error correction models (ECM) has this advantage: it describes the short-run dynamics of the variables as well as the long-run property of the model (see Bannerjee et al., 1993; Davidson and MacKinnon, 1993 and Verbeek, 2000; in Keele and DeBoef, 2004:9). Keele and DeBoef (2004) stated that estimating error correction model using the ARDL approach or the GECM approach yields consistent results. First, Keele and DeBoef specifies an ARDL (1,1) equation as follows:

[Y.sub.t] = [[alpha].sub.0] + [[alpha].sub.1][Y.sub.t-1] + [[beta].sub.0][X.sub.t] + [[beta].sub.1] [X.sub.t-1] + [[epsilon].sub.t] (1)

The Generalized Error Correction Model (GECM) was then derived as shown below:

[DELTA][Y.sub.t] = [[alpha].sub.0] + [gamma]([Y.sub.t-1] - [X.sub.t-1]) + [[lambda].sub.1][DELTA][X.sub.t] + [[lambda].sub.2][X.sub.t-1] + [[epsilon].sub.t] (2)

Where [gamma] = ([[alpha].sub.1] - 1), [[lambda].sub.1] = [[beta].sub.0] and [[lambda].sub.2] = [[beta].sub.1] + [[beta].sub.0] + [[alpha].sub.1] - 1

The GECM, unlike the ARDL model, directly tells how quickly the system reacts to any disequilibrium, as [gamma] the coefficient on the lag of Y is the error correction rate.

3. Analytical Framework

Annual time series data were collected from 1970 to 2010 from the Central Bank of Nigeria (CBN) 2010 statistical bulletin/website, the Nigerian National Bureau of statistics website and other relevant online sources. The theoretical framework adopted for the study is the Philips curve hypothesis (for the trade-off between unemployment and inflation), and the Okun's law (to examine the link between the Philips curve and economic growth). The econometric technique employed for the analyses is the generalized error correction model (GECM) adapted from Keele and DeBoef (2004) specified as follows:

[DELTA][UN.sub.t] = [[alpha].sub.0] + [gamma]([UN.sub.t-1] - [INF.sub.t-1]) + [[lambda].sub.1] [DELTA][INF.sub.t] + [[lambda].sub.2] [DELTA][INF.sub.t-1] + [[epsilon].sub.t] (3)

[DELTA][GDP.sub.t] = [[alpha].sub.0] + [gamma]([GDP.sub.t-1] - [UN.sub.t-1]) + [[lambda].sub.1] [DELTA][UN.sub.t] + [[lambda].sub.2] [DELTA][UN.sub.t-1] + [[epsilon].sub.t] (4)

[DELTA][DP.sub.t] = [[alpha].sub.0] + [gamma]([GDP.sub.t-1] - [INF.sub.t-1]) + [[lambda].sub.1] [DELTA][INF.sub.t] + [[lambda].sub.2] [DELTA][INF.sub.t-1] + [[epsilon].sub.t] (5)

Where UN is unemployment, INF is inflation; GDP is gross domestic product. [gamma], the coefficient on the lag of Y, is the error correction rate. It is the speed at which Y adjusts to any discrepancy between Y and X in the previous period. It has an a priori expectation of negative sign. The short-run effect is represented by [[lambda].sub.1]. [[lambda].sub.2] is a coefficient useful for obtaining the long run multiplier.

The data were logged and tested for stationarity before the OLS estimations were conducted. SPSS (statistical package for social sciences) was used to estimate the OLS linear and logarithmic regressions. The package also has the advantage of showing the scattered diagram with the linear and logarithmic plot in a diagram. The lagged regression models would also be estimated in an ARDL framework. This estimation is necessary to establish if the relationship connecting the variables follows a distributed autoregressive format. The lag length for each variable was determined using the Akaike Info Criterion (AIC). Eviews was used to estimate the model. The error correction models were estimated separately to establish the relationship between unemployment and inflation, unemployment and economic growth, and inflation and economic growth respectively. The models were also estimated using Eviews software.

4. Results and Discussion

The data were logged and then tested for stationarity. It was found that gross domestic product (GDP) and unemployment (UN) were while inflation (INF) was I(0). The summary results are shown in Table 4.1. Granger causality test was then conducted to determine the direction of causality. The result (which was significant at 10 percent level) showed that, INF granger causes UN. This implies that, inflation would be specified as the independent variable, while unemployment would be specified as the dependent variable. The granger causality result is presented in Table 4.2.

The scattered diagram showing the linear and logarithmic trend is presented in Figure 1 below.

[FIGURE 1 OMITTED]

The almost flat trend suggests an inverse but flat relationship between inflation and unemployment in Nigeria. The empirical test of the Philips curve hypothesis is estimated using the GECM. The estimated equation 3.1 is presented below.

Table 4.3 shows that the long run coefficient y, has the negative a priori sign and is significant. The short run coefficient [[lambda].sub.1], was negative but not significant. The short run coefficient is consistent with the findings from the OLS result implying that, the Philips curve hypothesis of a trade-off between inflation and unemployment in Nigeria is not significant in the short run. In the long run however, inflation and unemployment are in equilibrium implying that short run distortions between inflation and unemployment would be corrected in the long run making both variables to move in the same direction. The model generally has a good fit as the F-statistic probability value is significant at 5 percent and the Durbin-Watson statistic has an approximate value of 2. The coefficient [[lambda].sub.2] was also significant and had a negative sign. This implies that the GECM is consistent with the ARDL procedure and that inflation at lag one, has a significant negative relationship with unemployment. The adjusted r-square value shows that a dynamic change in inflation has a 15.09 percent effect on unemployment in Nigeria. The GECM result for GDP and Unemployment is presented below

As shown in Table 4.4, though the long run coefficient has the expected a priori sign, the coefficient was not significant. The short run coefficient was also not significant. The coefficient h2 was significant and had a negative sign. This implies that the GECM is consistent with the ARDL procedure and that unemployment at lag one has a significant negative relationship with gross domestic product. Generally, the result is significant at 10 percent and the Durbin-Watson statistics was also significant at a value of approximately 2. The adjusted r-square value shows that a dynamic change in unemployment has a 10.44 percent effect on gross domestic product. The GECM result showing the relationship between gross domestic product and inflation is presented in Table 4.5 below:

The long run coefficient [gamma], has the negative a priori sign but is not significant at 5 percent level. The short run coefficient [[lambda].sub.1], however was significant. The positive coefficient suggests that gross domestic product and inflation are positively related in the short run, which is also consistent with the OLS result. The model is generally significant at 10 percent level since the F-statistic has a probability value of 0.0988 and the Durbin-Watson statistic has an approximate value of 2. The coefficient [[lambda].sub.2], was also significant and had a positive sign. This implies that the GECM is consistent with the ARDL procedure and that inflation at lag one has a significant positive impact on gross domestic product. The adjusted r-square value shows that a dynamic change in inflation has a 9.03 percent effect on gross domestic product in Nigeria.

5. Summary and Policy Implication

The aim of this paper is to test the validity of the Philips curve hypothesis and to examine the implication of inflation and unemployment on the growth of the Nigerian economy. The generalized error correction model (GECM), was employed for the analysis. The evidence from the generalized error correction model (GECM) was estimated in three ways: (1) unemployment and inflation; (2) gross domestic product and unemployment; and (3) gross domestic product and inflation. For the first (unemployment and inflation), the result showed that though unemployment and inflation were negatively related in the short run, the result was not significant. The long run coefficient however shows that unemployment and inflation have long run equilibrium. This implies that unemployment and inflation move together in the long run and that short run distortions would be corrected in the long run, leading to stagflation. For the second case (gross domestic product and unemployment), the short run coefficient was negative implying that unemployment and GDP are negatively related. The result was however not significant. Also, though the long run coefficient had the a priori negative sign, it was also not significant. Lastly, the result for gross domestic product and inflation had a significant short run coefficient. This implies that GDP and Inflation are positively related in the short run. The result for the long run relationship was not significant. The policy implication is that current growth is accompanied by current inflation rates. However, growth is inversely related to lag levels of unemployment. Thus, inflation targeting should be pursued in a short time frame, to avoid a simultaneous case where unemployment and inflation accompanies growth. In conclusion, though the short run evidence of the inflation-unemployment relationship is weak, the long run evidence shows that, inflation and unemployment move in the same direction.

REFERENCES

Alogoskoufis, G., and Smith, R. (1991), "The Phillips Curve: The Persistence of Inflation and the Lucas Critique: Evidence from Exchange-Rate Regime," American Economic Review 54: 1254-1275.

Bhanthumnavin, K. (2002), "The Philips Curve in Thailand," paper presented at University of Oxford, June 13.

Bannerjee, A., D. Juan, W. G. John, and F. H. David (1993), Integration, Error Correction, and the Econometric Analysis of Non-Stationary Data. Oxford: Oxford University Press.

Davidson, R., and J. G. Mackinnon, (1993), Estimation and Inference in Econometrics. New York: Oxford University Press.

Garcia, C. J. (2010), "Is the Philips Curve Hypothesis Useful for Monetary Policy in Nigeria?" Occasional paper of the Research Department of the Central Bank of Nigeria No. 38, December.

Gokal, V., and S. Hanif, (2004), "Relationship between Inflation and Economic Growth," Working Paper 04 of the Economics Department, Reserve Bank of Fiji Suva.

Keele, L., and S. DeBoef, (2004), "Not Just for Cointegration: Error Correction Models with Stationary Data," University of Oxford, Working Paper 1213.

King, R. G., and M. W Watson (1994), "The Post-War U.S. Phillips Curve: A Revisionist Econometric History," Carnegie-Rochester Conference Series on Public Policy 41: 157-219.

Hansen, M., and R. Pancs, (2001), "The Latvian Labour Market Transition: the Beveridge and Phillips Curve as Indicators of Normalization," Euro Faculty, Riga.

Islam, F., K. Hassan, M. Mustafa, and M. Rahman (2003), "The Empirics of U.S. Phillips Curve: A Revisit," American Business Review 20(1): 107-112.

Moosa, I. A. (2008), "Economic Growth and Unemployment in Arab Countries: Is Okun's Law Valid?" Paper presented at the International Conference The Unemployment Crisis in the Arab Countries, 17-18 March, Cairo.

NPC (2004), "Meeting Everyone's Needs: National Economic Empowerment and Development Strategy," Nigerian National Planning Commission (NPC), Abuja.

NPC (2007), "National Economic Empowerment and Development Strategy 2 (NEEDS II)," Nigerian National Planning Commission (NPC), Abuja.

Ogbokor, C. A. (2005), "The Applicability of the Short-run Phillips Curve to Namibia," Journal of Social Sciences 1(4): 243-245.

Roeger, W., and B. Herz (2012), "Traditional versus New Keynesian Phillips Curves: Evidence from Output Effects," International Journal of Central Banking 8(2): 87-109.

Rudd, J., and K. Whelan (2006), "Can Rational Expectations Sticky-Price Models Explain Inflation Dynamics?" American Economic Review 96(1): 303-320.

Verbeek, M. (2000), A Guide to Modern Econometrics. New York: John Wiley and Sons.

KANAYO OGUJIUBA

kannyog@gmail.com

Department of Statistics and Population Studies

University of Western Cape

TERFA W. ABRAHAM

Lorenzcurve@yahoo.com

National Institute for Legislative Studies, Abuja
Table 4.1: STATIONARITY TEST: Augmented Dickey
Fuller (ADF) Unit Root Test

Level test I(0)                Critical Values

ADF Stat          Variables    1%        5%        10%

-0.1642           GDP         -3.6067   -2.9378   -2.6069
-2.7047           UN          -3.6067   -2.9378   -2.6069
-3.945902         INF         -3.6067   -2.9378   -2.6069

First Difference Test

-5.3123           GDP         -3.6117   -2.9399   -2.6080
-9.3197           UN          -3.6117   -2.9399   -2.6080

Source: Author(s) Presentation from Eviews Output

Table 4.2: Pairwise Granger Causality Tests For UN and INF

Date: 04/29/11  Time: 11:18 Sample: 1970 2009
Lags: 2

Null Hypothesis:                 Obs   F-Statistic   Probability

INF does not Granger Cause UN    38    2.65094       0.08558
UN does not Granger Cause INF          0.12541       0.88256

Source: Author(s) Estimation from Eviews

Table 4.3: GECM Result for Equation (3)

Dependent Variable: [DELTA][UN.sub.t]
Method: Least Squares
Sample(adjusted): 1971 2010
Included observations: 39 after adjusting endpoints

Variable                                Coefficient  Std. Error

[[alpha].sub.0]                          0.342355    0.113769
[gamma]([UN.sub.t-1] - [INF.sub.t-1])   -0.337427    0.116812
[[lambda].sub.1][DELTA][INF.sub.t]      -0.053942    0.052568
[[lambda].sub.2]][INF.sub.t-1]          -0.446532    0.143947

Variable                                t-Statistic  Prob.

[[alpha].sub.0]                          3.009218    0.0048
[gamma]([UN.sub.t-1] - [INF.sub.t-1])   -2.888639    0.0066
[[lambda].sub.1][DELTA][INF.sub.t]      -1.026148    0.3119
[[lambda].sub.2]][INF.sub.t-1]          -3.102056    0.0038

R-squared             0.217895    Mean dependent var       0.001455
Adjusted R-squared    0.150858    S.D. dependent var       0.103733
S.E. of regression    0.095589    Akaike info criterion   -1.76060
Sum squared resid     0.319805    Schwarz criterion       -1.589978
Log likelihood       38.33169     F-statistic              3.250347
Durbin-Watson stat    2.261764    Prob(F-statistic)        0.033243

Source: Author(s) Estimation from Eviews Output

Table 4.4: GECM Result for Equation (4)

Dependent Variable: [DELTA][GDP.sub.t]
Method: Least Squares
Sample(adjusted): 1971 2010
Included observations: 39 after adjusting endpoints

Variable                                Coefficient   Std. Error

[[alpha].sub.0]                          0.328320     0.104664
[gamma]([GDP.sub.t-1] - [UN.sub.t-1])   -0.012205     0.011582
[[lambda].sub.1][DELTA][UN.sub.t]       -0.035485     0.128606
[[lambda].sub.2][UN.sub.t-1]            -0.275615     0.105106

Variable                                t-Statistic   Prob.

[[alpha].sub.0]                          3.136884     0.0035
[gamma]([GDP.sub.t-1] - [UN.sub.t-1])   -1.053772     0.2992
[[lambda].sub.1][DELTA][UN.sub.t]       -0.275924     0.7842
[[lambda].sub.2][UN.sub.t-1]            -2.622247     0.0128

R-squared             0.175111   Mean dependent var       0.094108
Adjusted R-squared    0.104406   S.D. dependent var       0.080542
S.E. of regression    0.076221   Akaike info criterion   -2.213435
Sum squared resid     0.203340   Schwarz criterion       -2.042813
Log likelihood        47.16197   F-statistic              2.476645
Durbin-Watson stat    1.949407   Prob(F-statistic)        0.077474

Source: Author(s) Estimation from Eviews Output

Table 4.5: GECM Result for Equation (5)

Dependent Variable: [DELTA][GDP.sub.t]
Method: Least Squares
Sample(adjusted): 1971 2010
Included observations: 39 after adjusting endpoints

Variable                                 Coefficient   Std. Error

[[alpha].sub.0]                          -0.017375     0.080988
[gamma]([GDP.sub.t-1] - [INF.sub.t-1])   -0.000917     0.011033
[[lambda].sub.1][DELTA][INF.sub.t]        0.094718     0.042238
[[lambda].sub.2][INF.sub.t-1]             0.099261     0.045439

                                         t-Statistic   Prob.

[[alpha].sub.0]                          -0.214531     0.8314
[gamma]([GDP.sub.t-1] - [INF.sub.t-1])   -0.083160     0.9342
[[lambda].sub.1][DELTA][INF.sub.t]        2.242465     0.0314
[[lambda].sub.2][INF.sub.t-1]             2.184488     0.0357

R-squared             0.162121     Mean dependent var       0.094108
Adjusted R-squared    0.090302     S.D. dependent var       0.080542
S.E. of regression    0.076819     Akaike info criterion   -2.197810
Sum squared resid     0.206542     Schwarz criterion       -2.027188
Log likelihood       46.85729      F-statistic              2.257374
Durbin-Watson stat    2.110489     Prob(F-statistic)        0.098887

Source: Author(s) Estimation from Eviews Output
COPYRIGHT 2013 Addleton Academic Publishers
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2013 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Ogujiuba, Kanayo; Abraham, Terfa W.
Publication:Economics, Management, and Financial Markets
Article Type:Report
Geographic Code:6NIGR
Date:Dec 1, 2013
Words:3704
Previous Article:Globalization and developing countries: a global participation model.
Next Article:Typologies and sources of entrepreneurial opportunity (II).
Topics:

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters