Recent macroeconomic performance in Colombia: what went wrong?

• Author: Arango, Luis E.
• Pages: 1(19)

Resumen. Al finalizar la década anterior la actividad real en Colombia experimentó la más aguda recesión de los últimos 50 años. Para explicar este fenómeno, postulamos un modelo VAR estructural no-triangular que describe la dinámica de la producción, los precios, el desempleo y los salarios durante las últimas dos décadas. La evidencia sugiere que, en el largo plazo, la política monetaria ha sido neutral con respecto al producto y al desempleo, mientras que la principal razón para el incremento de este último se explica por la forma en que se han determinado los salarios (formación de expectativas hacía atrás) y el incremento de los costos no salariales.

Palabras clave: VAR estructural, desempleo, política monetaria, salarios, costos no-salariales del trabajo, expectativas.

Clasificación JEL: E24, C40.

Abstract. At the end of the last decade, real activity in Colombia underwent the sharpest recession it had suffered in the last fifty years. With the aim to explain this phenomenon, we are postulating a non-triangular structural VAR model to describe the dynamics of output, prices, unemployment and wages during the last two decades. Evidence suggests that, in the long run, monetary policy has been neutral in regard to both output and unemployment while the main reasons for the increase in the latter have been the way in which wages have been determined (because of backward-formed expectations) and the increase in non-wage labour costs.

Key words: structural VAR, unemployment, monetary policy, wages, non-wage labour costs, expectations.

JEL classification: E24, C40.

1. Introduction

   At the end of the 1990's, Colombian economy suffered the most serious recession in its last fifty years. It was so deep that output decreased about 5% in 1999. In addition, the unemployment rate started to rise consistently and reached 20% in 2000. This increase in unemployment was accompanied by a gradual reduction in inflation and an increase in real wages.

   Several reasons have been put forward to explain the slowdown of the economic activity. First, the tight monetary policy set by an authority committed to an inflation reduction program. Second, the increase in the cost of labour input due to three elements: (1) the type of expectations formed by agents when setting nominal wages, which pushed wages above their long run equilibrium level, (2) the minimum wage policy, and (3) the increase in non-wage labour costs (Arango and Posada, 2001, 2002). Urrutia (2002) offers another explanation for the recession of 1999 which is linked to the deficit of the current account and the sudden stop of capital flows associated with the international capital markets crisis which occurred in 1997 (see also Izquierdo and Galindo, 2003). However, this view is not emphasized in our work since we will first intend to understand the internal causes of the crisis. A natural step to take towards that goal is to consider external aspects of that crisis.

   Accordingly, we postulate a structural VAR model for a closed economy to describe the dynamics of output, unemployment, wages and prices during the last two decades. This type of approach has been previously used to study several macroeconomic aspects in different economies. For example, Dolado and Jimeno (1997) associate the causes of unemployment in Spain with different types of shocks which have long lasting effects due to a full hysteresis phenomenon. They find that the "dismal performance of Spanish unemployment can be explained as the result of a series of adverse shocks, which were difficult to absorb in a context of a rigid system of labour market institutions and disinflationary policies. This finding has relevant policy implications suggesting that, unless supply side reforms are implemented, deflationary policies will continue to be very costly in unemployment terms" (p. 1285).

   Balmaseda et al. (2000) focus on the role played by aggregate demand, productivity, and labour supply shocks in explaining the joint dynamic behavior of real output, real wages and the unemployment rate in a model for the labour markets of sixteen OECD economies over the period 1950-1996. They find that "in most countries the identification scheme based on unemployment being persistent but stationary yields more reasonable results than those based on full-hysteresis whereby unemployment is considered to be an I(1) variable" (p. 22). In addition, the authors find that unemployment fluctuations are dominated by aggregate demand shocks in the short run and by labour supply and productivity shocks at lower frequencies. (See also Algan, 2001 and Fabiani et al., 2001.)

   In the case of Colombia, Misas and Lopez (1998, 2001) use a SVAR model along the lines of Blanchard and Quah (1989) to estimate output and unemployment gaps. Misas and Posada (2000) examine the sources of variation in the unexpected component of output growth. Arango (1998), also with a SVAR approach, provides some evidence on the nature (either nominal or real) of the temporary and permanent components of Colombian output and prices. Finally, Restrepo (1997) uses a VAR approach to explain the response of some macroeconomic variables (GDP, real exchange rate index, real money balances, money, inflation, and interest rate) to demand and supply shocks in the markets for goods and money.

   Unlike existing literature on Colombian economy, in this paper we present and solve a stylized model to study the economy activity's recent macroeconomic behavior. An appealing of this work is that the empirical approach introduces the long-run restrictions provided by the theoretical framework. The evidence we provide suggests that, in the long run, monetary policy has been neutral as far as both output and unemployment are concerned. The main reasons for the increase in the latter have been the way in which wages are determined (which increased the real wage), and the rise in non-wage labour costs, such as those introduced by pension reforms. (1)

   The outline of this paper goes as follows. Some facts related to the macroeconomic performance of the Colombian economy during the last few years are presented in Section 2. The model is introduced in Section 3. The methodology and the empirical application are presented in Section 4. A discussion of our findings is reported in Section 5 and some concluding remarks are provided in Section 6.

2. The facts

   Since 1991, Colombia's central bank has been conducting a program that is intended to reduce inflation. This program has been characterized by targets that decrease gradually, accompanied by, among other things, consistent stances of monetary policy. (2) On one hand, the policy was effective in the sense that, since that year, inflation has shown a negative-sloped long-run component (see Figure 1). On the other hand, the program for inflation reduction was not as successful as expected since no inflation target was reached until 1997 (see Figure 1). In 1997 the inflation target was 18 percent while the observed inflation was 17.7 percent. In 1998, the authorities missed the target again. In 1999 and 2000, the observed inflation levels were 9.2 percent and 8.8 percent while targets were 15 and 10 percent, respectively. These results may have undermined the credibility of the monetary authority and, ex post, might suggest that the monetary policy stance was tighter than necessary. (3)

   [FIGURE 1 OMITTED]

   To complete the picture, in the late 1990's, Colombia experienced a deep recession as shown in panels A, B, and C of Figure 2. The growth of real...