CFGE Simulations and Analysis

To the extent that certain forms of relations (shape of the curve) have different impacts on poverty when analyzed under expansionary and contractionary policy, the following analysis was conducted under these two sets of simulation. In each set I distinguished the impact of monetary (interest rate) and fiscal (government expenditure) policy. The mechanisms to arrive at the poverty condition are based on the CFGE model, and the focus is on the relations among four variables in the 4-quadrant setting described earlier. The ultimate variable of interest is the real income of the poor denoted by Y^Poor (deflated by the appropriate poverty line prices). Since the role of income inequality in affecting the growthpoverty nexus is critical, the simulations in each scenario also include relative income distribution.

From the sensitivity analysis and re-examination of the model structure, it was revealed that the wage equation, particularly parameter ρ that links WAGES and PINDEX (in the equation below)

holds the key to the results; influencing not only the factor income but also prices. More fundamentally, while incomes of the poor change according to the policy shock, i.e., declines under a contractionary policy and increases when the economy expands, the effect on real income of the poor Y^Poor depends critically on what happens with the prices and the poverty line. In scenarios where ρ are low, the rate of income change is faster than the rate of price change, while in large ρ scenarios the reverse applies. Thus, applying different values of ρ in each simulation is warranted.

Under low ρ, an expansionary monetary policy (by lowering the interest rates successively) in Indonesia would generate higher income for the poor but it would also create a higher poverty line threshold (hence higher PL). On the other hand, higher interest rates would lower the income of the poor and PL. The net effect on Y^Poor was eventually determined by which of the two changes is larger. As shown by the curve to the right of the “Base” point in Figure 14 [ PDF 57.3KB | 1 page ], lowering interest rates successively would likely raise the Y^Poor, implying that the rate of income increase is higher than the rate of PL increase. Conversely, by raising the interest rates successively, the Y^Poor tends to decline (left side of the vertical line), suggesting that income will fall faster than PL.⁹ Note, however, that the change in Y^Poor in both directions is at a decelerating rate. This is to be expected: given the prevailing excess capacity, a positive shock of AD would effectively raise GDP without a strong inflationary pressure. But as the excess gets smaller and eventually disappears, a further positive shock of AD would become less effective in stimulating growth and would generate strong inflationary pressure, such that the expected rate of increase of GDP and real income of the poor would decelerate.

Less expected is the outcome of the scenario under fiscal policy simulation. As shown in Figure 15 [ PDF 60.1KB | 1 page ], higher government expenditure tends to reduce real income of the poor. Two possibilities may explain this result. First, the expenditure shock is applied without making any changes in the sectoral composition of expenditure. The fact that the base year of the financial social accounting matrix (FSAM) used in the model is 2005, during which a less pro-poor policy took place (there were two drastic cuts on the subsidy for domestic fuel price—one in March and another in October of 2005 (Azis, 2006)), implies that the successive shocks in the simulation continue to be less favorable for low income households. Secondly, the financial channel tends to work to the benefit of high income households because interest rates and returns on financial assets increase as a result of the rightward shift of the IS curve. In turn, this generates extra earnings for high income households who own such assets, but it also drives up prices. While incomes of the poor are unaffected, the resulting poverty line tends to increase, thereby lowering the Y^Poor.

The simulation results based on Thailand's CFGE model, however, show a different pattern. A monetary expansion would have caused higher inflation and poverty line in such that the Y^Poor tends to decline (Figure 16 [ PDF 60.1KB | 1 page ]). But a fiscal expansion is likely to raise the income of the poor more than the price increase. As shown in Figure 17 [ PDF 59KB | 1 page ], given a low ρ the greater the fiscal spending the higher the Y^Poor. The non-linear nature of the relations is clearly shown by the accelerated increase and the decelerated decrease of Y^Poor under the higher and lower spending conditions, respectively. The turning point under lower spending is at around 50%; a further cut beyond that level would result in a sharp fall of the poverty line such that the Y^Poor tends to increase.¹⁰

What happens with the resulting income inequality under the above policy scenarios? The favorable effect of monetary expansion on poverty in Indonesia tends to reduce income inequality as indicated by the dynamic trend of the slope between GDP and income of the poor (see the relation in quadrant 4 of Figure 12). As the interest rates are lowered, the slope gets larger, suggesting that the income of the poor increases more proportionally than the increase of GDP. On the other hand, a contractionary monetary policy tends to worsen income inequality as indicated by the decreasing size of the slope to the left of the vertical line in Figure 18 [ PDF 59KB | 1 page ]. This is similar to the case in Thailand. Although an expansionary policy through successive falls of interest rates would have generated lower Y^Poor, the resulting income distribution would have slightly improved, as indicated by the increasing size (decreasing negative value) of the slope as shown in Figure 19 [ PDF 58.8KB | 1 page ].

A further look at the mechanisms points to the vital role of price (hence of the PL). In Thailand's case, the inflationary pressure (rising PINDEX) will not cause a price-spiraling effect; this is because a low ρ implies that price changes have only a mild effect on wages. With relatively smaller inflationary pressure and lower increase of PL, the effect of income deflator is also limited, leading to rising real income of the poor. On the other hand, in the Indonesian case, the effect of monetary policy on price level is larger, so the increase of incomes of the poor is likely offset by the PL increase. On relative income distribution, the channel through which income from financial assets is determined turns out to play an important role. As the returns on financial assets fall with interest rates, the incomes of those owning such assets (high income groups) also declines. Poor households, on the other hand, are unaffected; they are insulated from the changing returns on financial assets as they generally do not hold such assets. This finding highlights the importance of incorporating the financial sector in the model.

The distributional effect of fiscal expansion in Thailand is expected: it tends to lower income inequality as indicated by the rising slopes of GDP and Y^Poor in Figure 20 [ PDF 58.8KB | 1 page ]. Less expected are the results for Indonesia: if fiscal spending is raised, the income distribution tends to get worse as the slope tends to get smaller. On the other hand, if the spending is lowered, the income distribution is likely to get better as the slope tends to get larger) (Figure 21 [ PDF 58.4KB | 1 page ]). As argued earlier, this is likely due to the fact that the expenditure composition in 2005, the base year used in the model was less pro-poor and that the fiscal expansion assumed in the scenario keeps such a composition unchanged.

When ρ are higher, the results of model simulations for Indonesia show that the increase of income following an expansionary policy is less than the increase of price and PL. Consequently, lowering interest rates worsens the poverty condition and real income of the poor tends to decline (Figure 22 [ PDF 58.4KB | 1 page ]). On the other hand, a monetary tightening through successive interest rate increases tends to improve the poverty condition as the decrease of PL is larger than the fall in income. A similar pattern is observed when the expansion and contraction are conducted through fiscal policy (Figure 23 [ PDF 66.8KB | 1 page ]). The decline in Y^Poor following increased expenditure, however, occurs at a more decelerated rate.

In the Thailand case, the size of ρ does not seem to influence the adverse effect of a positive monetary shock on poverty: Y^Poor tends to decline, just as in the case of low ρ (Figure 24 [ PDF 66.8KB | 1 page ]).

Interestingly, the effect of fiscal expansion is also unaffected by the size of ρ; it continues to be favorable for poverty reduction (Figure 25 [ PDF 66.8KB | 1 page ]). Thus, it is clear that as far as the effect of macroeconomic policy on poverty is concerned, the size of ρ matters in Indonesia but not in Thailand. A fiscal expansion in Thailand is favorable for poverty reduction, whereas a positive monetary shock in Indonesia is favorable when the price elasticity of wages is low.

To the extent that the fiscal policy in Thailand during the post-crisis period was actually expansionary (discussed in Section II), a poverty decline should be expected. As shown in Figure 26 [ PDF 54KB | 1 page ] and Figure 27 [ PDF 54KB | 1 page ], this was indeed the case since 2000. What is also obvious is that since 2000, the trend in the two countries has differed: the poverty decline has been consistent in Thailand but has fluctuated in Indonesia. In fact, during 2004–2006 the number of poor in Indonesia has increased; there were more poor people in 2006 than in 1996, the year before the crisis. Such a disappointing trend is consistent with the absence of expansionary AD policy that has led to meager growth performance, despite the fact that the country's AS curve is flat. As shown earlier, under certain conditions (i.e., low ρ) a more monetary expansion would have been able to reduce poverty.

The importance of output growth in generating higher incomes of the poor is also supported by the fact that the increasing trend of the poverty line in Thailand has been more significant than in Indonesia (Figure 28 [ PDF 59.7KB | 1 page ]), yet the poverty incidence has been favorable in the former but not in the latter.

The resulting income distribution of monetary expansion with a larger ρ is favorable in Thailand, as indicated by the increasing slope of GDP-Y^Poor (Figure 29 [ PDF 59.7KB | 1 page ]). Thus, as far as the impact of monetary policy is concerned, the size of ρ in Thailand influences neither the impact on poverty nor on income inequality. The same is true with respect to the impact of fiscal policy (Figure 30 [ PDF 59.7KB | 1 page ]). In the Indonesian case, although the size of price-wage elasticity can reverse the resulting poverty of monetary policy (discussed earlier in this section), it does not really affect the resulting income inequality. As shown in Figure 31 [ PDF 61KB | 1 page ] and Figure 32 [ PDF 61KB | 1 page ], the slope under a high ρ scenario tends to decline as is also the case when ρ is low (Figures 18 and 20).

Download this Discussion Paper [ PDF 978KB| 34 pages ].

Post a Comment

We welcome your feedback on this publication. Post a comment. ADBI is not obliged to acknowledge or publish comments and may abridge or edit them before web posting.

Comment(s)

There are [1] comment(s) for this entry. Post a comment.

1. Ruly
   (posted 11 July 2008 / 02:08:15 PM)
   
   This paper gives me a new insight about the real condition of my country economy, and how the authority (government) had applied many conventional economic policies, which is not suitable for Indonesian economy that has different characters and conditions from the west, the origin of these conventional policies..
   
   Furthermore, I was very lucky because the author presented this paper in my class as a guess lecturer about 2 months ago. For the first time after 1.5 years studying economics, a lot of my "unanswered" questions being answered, and I even didn't have to ask..