Contact

Understanding the policy position of the South African Reserve Bank – a structuralist trilemma

Michèlè Capazario

There is heated debate in South Africa at the moment surrounding the mandate of the South African Reserve Bank (SARB). Currently, its mandate revolves around maintaining inflation- the growth rate of market prices- within a particular band of between 3 and 6% (South African Reserve Bank, 2019). This is generally known as “inflation targeting”.

Rhetoric in the political space currently suggests that this mandate should also include,inter alia, room for “growth targeting”. That is, instead of solely influencing inflation with changes in the interest rate, there should be a move towards influencing our country’s Gross Domestic Product as well.

Importantly, and I say this as a disclaimer, this piece does not express an opinion on the mandate of the Reserve Bank; instead, my goal is to discuss the function of the SARB and its possible economic impact, supported by some basic analysis. This will be done using a Structuralist lens- a way to analyse the economy with a broad focus on how effective demand within an economy is defined, particularly (and this is the distinction between more mainstream frameworks) paying attention to income distribution dynamics, and how they may influence the macroeconomy.

Before this discussion can take place, however, it is important to understand the avenue through which the SARB is able to influence the economy- through changes in the lending/interest rate.

For instance, assuming that the SARB decided to hike interest rates, one would expect two things to take place in an open economy:

        1. Interest sensitive expenditure would decrease- this implies that consumption and investment based on loans (which are now more costly to service due to the increased lending rate) would decline (Blanchard, 2014), and
        2. The exchange rate would (theoretically, at least) appreciate- given that the South African interest rate might be higher than foreign interest rates, this also poses a potential gain for foreign investors, who might want to put their money into South African bank accounts and money market instruments. This inflow of foreign currency tends to prop up our own, relatively appreciating the Rand (Appleyard, et al., 2008). This is caveated by the fact that, when looking at a depreciation/appreciation in the South African currency, it is perhaps only a real depreciation/appreciation which impacts output growth in the longer term; literature tends to suggest that output growth is only really influenced in the short term by changes in the nominal exchange rate (Edwards & Garlick, 2008).
           
Nevertheless, because the Rand would have appreciated, importing goods from overseas would be relatively cheaper. Exporting to other countries, however, would become relatively more expensive. This depresses net exports (especially if this appreciation is real, rather than nominal). Furthermore, because consumption and investment both drop off, we see a general decrease in our country’s aggregate demand (a proxy for our nation’s Gross Domestic Product). This fall in aggregate demand decreases the price level within our country, andipso facto, this keeps inflation in check (Blanchard, 2014). Of course, if the inverse were to happen, and the interest rate were to go down, this would have the opposite effect on interest sensitive expenditure and the exchange rate- increasing aggregate demand and pulling up inflation.
 

This creates a trade-off/dilemma for the SARB- should they increase the interest rate so as to keep prices in check, but ultimately decrease short term aggregate demand; or, should interest rates be cut, so as to stimulate short term economic growth, but in-so doing, increase inflation? This inflation-output trade-off is best understood by using a simple, yet elegant tool- the Taylor Rule.

In simple terms, the Taylor Rule- developed by John Taylor in 1993- relates how the interest rate adjusts when output and inflation change within an economy. This helps us to understand some of the key variables that influence changes in interest rates, though it certainly does not incorporate all of the factors that inform the actual decisions of the SARB. As Mohanty & Klau (2005) point out, this is particularly relevant for emerging markets whose monetary policy decisions are paramount for macroeconomic stabilization.

As such, I estimated two (very simple) South African Taylor Rules- one, quite standard, and the other, slightly less so. The first is of the functional form[1]:

it = αc+ αiit-1+ απ(πt- πt) + αygyt

The equation simply says that the current short-term (nominal) interest rate,it will relate positively to the previous quarter’s short-term interest rate, it-1,the deviation of inflation(πt)from its target(πt)and annualized output growth (gyt).

More descriptively, if a previous quarter’s interest rates were set at a particular level, the inertia from that interest rate would carry over into the current period (in order for interest rate volatility to be smoothed out), hence αi is expected to be positive and close to 1 (Rudebusch & Svensson,1999).

 

Furthermore, if inflation is above its target, it stands to reason that an inflation-targeting reserve bank would want to hike interest rates, and in so doing pull inflation down towards its target. Hence, if inflation exceeds its target level, the interest rate would most likely increase by απ% (Taylor, 1993).
 
Finally, focusing on output growth- I have discussed before that increased aggregate demand usually leads to increased prices, which then leads to inflationary pressure. Given a percentage increase in output growth, the SARB would most likely increase the interest rate by a factor of αy%(Taylor, 1993), in order to keep inflation in check.

 

This simple rule, however, does not consider one very important detail – how the central bank reacts to changes in income distribution . Whereas certain Taylor rules may also include exchange rate or money supply terms (Mohanty & Klau, 2005), it is only in the Structuralist literature that a proxy for income distribution is used as a means to explain monetary policy decisions

As Taylor (1991) points out, changes in the distribution of wealth between classes has a large impact on inflation (due, in part, to wage changes, and in part to changes in consumption behaviour), and as such, it is within the mandate of the SARB to focus on possible changes to income distribution.

More formally, given the imperfect data world in which we live, a proxy for income distribution that is used in Structuralist modelling this is the labour share[2] (i.e., the proportion of national income that is earned by workers). For context though, an economy can either be wage-led (where output growth positively relates to increasing the labour share), or profit-led (where output growth negatively relates to increasing the labour share) (Taylor, 2009). South Africa is profit-led (Malikane, 2017), signalling that as the labour share rises, output growth shrinks- this then leads to unemployment, which ultimately cuts into the labour share, bringing about a less equal income distribution, whilst also (theoretically) leading to decreased inflation.

Therefore, instead of output growth (due to possible endogeneity and multicollinearity issues- for a breakdown of what these are, see Gujarati & Porter (2009), I use the growth in the labour share to come up with a more nuanced Taylor Rule[3]of the following form:

it = αc+ αiit-1+ απ(πt- πt) + αs(st- st-20)

In this variant, it is important to note that changes in income distribution take time to influence the economy (inflation, particularly)- hence, the SARB would only react to income distribution changes after a relatively long period of time. For this reason, the labour share variable,(st– st-20), measures the change in income distribution over the previous 20 quarter/5-year period.

In line with Structuralist literature then, the relationship between the interest rate and labour share growth is ambiguous. It is the inherent structure of the economy within which the Taylor Rule is estimated that defines whether there is a positive or negative relationship between the two.

If, for instance, the labour share was to increase, there are two ways in which this change could influence inflation, either:

        1. The inflation regime is reformist- in the first round, an increase in wage demands would drive up the labour share to a certain degree. This increased wage demand comes about due to inflationary pressure in the country. Due to the inherent structure of the labour market (particularly, due to the large number of unemployed reserve workers in the labour market who would be able to replace current workers at a lower wage), however, this increase in wages would likely lead to a rise in unemployment in the second round (Taylor, 1991). This increased unemployment cuts into consumption spending in the country, decreasing output, and with it, decreasing inflation (Phillips, 1958); or
        2. The inflation regime is militant- in the first round, an increase in wage demands would drive up the labour share to a certain degree. In this case, however, there is no large reserve army of workers to replace current workers. Instead of retrenching workers who ask for higher wages, employers acquiesce to this wage demand, and workers in the economy are now able to increase their consumption due to their improved wealth (Taylor, 1991). This bids down unemployment in the second round due to increased national output, and with it, inflation increases (Phillips, 1958).

 
Thus, an increase in the labour share would create inflation in a militant regime, and the SARB would most likely increase the interest rate to combat this. If, however, the inflation regime is reformist, an increase in the labour share could see a future dip in South African inflation, which the SARB would look to offset through a decrease in the interest rate.

This explanation, admittedly, abstracts from the slope of the Aggregate Supply (AS)/output-based Phillips Curve (i.e., how elastic or inelastic the relationship is between output and inflation is, or even whether there is non-linearity in this relationship (Akerlof, et al., 2001). Very simply, a very elastic AS (meaning that a large change in output yields a relatively small change in inflation) curve could imply that a militant regime would see only a small inflationary trade-off if the country were wage-led.

Given that in South Africa:

    1. The unemployment rate is so high,
    2. The labour force is not highly skilled, and
    3. Education quality is relatively poor

It stands to reason that the South African inflation regime is reformist with relatively limited ability to improve wages while keeping unemployment constant (here, see a discussion in Faulkner, et al., (2013) which discusses the inherent structure of the South African employment landscape, inter alia).

In the South African case then, being both reformist and profit-led, an increase in the labour share would see two things:

        1. A decrease in output growth, and
        2. A drop in the inflation rate.

Along an elastic AS, the change in output would be far greater than the drop-in inflation, while the inverse holds true for an inelastic AS curve[4]. In line with this all, the results of these two estimations (using South African data from between Q1 of 1994 and Q4 of 2018) are shown in the table below:


In equation 1, we see that all signs and coefficients are as we would expect. If the previous quarter’s short-term interest rate was 1%, the current interest rate would be 0.91%. Put differently, approximately 91% of changes in the current interest rate are driven by changes in the previous interest rate. Furthermore, if inflation was to increase by 1% above its target level, the central bank would likely hike the interest rate by 0.14%[5]in order to curb the ensuing inflationary pressure. Similar coefficients are reported for the second equation’s previous interest rate and inflation terms.

 

Most importantly for our discussion though, is the inclusion of output growth in equation 1, and labour share growth in equation 2.

Given equation 1, I find that an increase in 4-quarter annualized output growth sees an increase in the interest rate of approximately 0.13%. This itself is a growth retardant policy, aimed at decreasing output so as to keep inflation in check.

 

Moving focus to equation 2, an increase in the labour share relates to a decrease in the interest rate. This confirms 2 things:

        1. The South African inflation regime is most likely reformist- where an increased labour share would probably lead to an increase in unemployment and decrease inflation in the short term as previously described, and
        2. Because of this, the central bank looks to stimulate demand by dropping the interest rate- hence the negative relationship as outlined in the table.

Thus, an increase in the labour share by 1% above its 5-year-previous level would see a decrease in the interest rate of around 0.13%. Whereas this is itself a relatively growth friendly policy decision from the perspective of output, it is important to note the longer-term impact that this may have on income distribution.

The labour share is defined as the country’s wage bill divided by price-normalized national income[6](i.e.st= WtLt⁄ PtYt). It therefore stands to reason that the inflation caused by the ensuing economic expansion will cut into the labour share (as prices and output rise, the labour share shrinks mathematically). This is due, in part, to the diminished purchasing power of the working class (given price hikes), and in part, due to the way in which income growth benefits are shared across classes in the country (given that the economy is also profit-led, output growth and profits are positively related (Malikane, 2017), meaning that those who tend to enjoy output growth benefits the most are the capitalist class).

Thus, even though a growth-friendly policy is expansionary in output, it will likely have a negative impact on the share of national income which workers in the South African economy receive.

Based on this analysis, monetary policy can either be:

      1. Inflation-targeting, which can depress output in the short-term when interest rates are rising, but maintains levels of income distribution between classes, or 
      2. Growth-targeting, which can improve short term demand growth, but tends to diminish labours’ share in national income, skewing the distribution of income even further towards the capitalist class.

The obvious trade-off faced by the reserve bank is then whether to support higher short-term growth (at the risk of higher inflation), or to pursue price stability and a relatively more equal income distribution (at the cost of higher short-term interest rates). This SARB trilemma (between output, inflation and income distribution), instead of the regularly thought about dilemma (between output and inflation), is something that needs further discussion and dissection, and something that I believe needs to come to the fore in research to come.


[1] In the Taylor (1993) paper, the output gap is used. As per Hamilton (2018), however, it is important to note that the estimation of potential output may inherently bias the modelling process, due to the imprecise use of various filters and statistical techniques. As such, output growth is perhaps a more reliable measure to include, without structurally deriving potential output.

[2] For a more full description of this, please see the linked article-https://www.dnaeconomics.com/pages/public_finance/?zDispID=NewsArtProfitLed_or_WageLed_Growth_How_Income_Distribution_Influences_the_Macroe

[3] This form of Taylor rule can be traced back to works by Flaschel & Krolzig (2002), although it is most prominently featured in Taylor (2009).

[4] Furthermore, and if, in fact, the South African AS curve were non-linear in the way that Akerlof, et al. (2001) describe, although in the inflation-output space, this may see less pronounced declines in output growth or even an output growth increase for a given inflation band around which the non-linearity/retrovertion is most pronounced.

[5] This reaction is seemingly small, but is due to the inclusion of the lagged interest rate term in the Taylor Rule. The inclusion of this term, apart from bolstering the robustness of the rule from a statistical perspective, also exerts downward pressure on the size of the coefficient attached to inflation (Rudebusch & Svensson, 1999).

[6] Wt– average wages; Lt –the number of working class employed in the economy; Yt – output/national income; Pt – price level in the economy.

Bibliography

Akerlof, G., Dickens, W. & Perry, G., 2001. Options for Stabilization Policy. The Brookings Institution Policy Brief, Volume 69, pp. 1-8.

Appleyard, D., Field, A. & Cobb, S., 2008. International Economics. 6 ed. s.l.:s.n.

Blanchard, O., 2014. Macroeconomics. A South African Perspective ed. s.l.:Pearson Higher Education.

Edwards, L. & Garlick, R., 2008. Trade Flows and the Exchange Rate in South Africa, Cape Town: University of Cape Town.

Faulkner, D., Loewald, C. & Makrelov, K., 2013. Achieving Higher Growth and Employment: Policy Options for South Africa. South African Reserve Bank Working Paper Series, 13(03), pp. 1-34.

Flaschel, P. & Krolzig, H.-M., 2002. Wage and Price Phillips Curves. CEM Working Paper Series, Volume Working Paper 66.

Gujarati, D. & Porter, D., 2009. Basic Econometrics. International Edition ed. s.l.:McGraw-Hill .

Hamilton, J., 2018. Why You Should Never Use the Hodrick-Prescott Filter. Review of Economics and Statistics, 100(5), pp. 831-843.

Malikane, C., 2017. The Labour Share and the Dynamics of Output. Applied Econometrics, 49(37), pp. 3741-3750.

Mohanty, M. & Klau, M., 2005. Monetary Policy Rules in Emerging Market Economies: Issues and Evidence. In: Monetary Policy and Macroeconomic Stabilization in Latin America. Berlin: s.n., pp. 205-245.

Phillips, A., 1958. The Relation Between Unemployment and the Rate of Change of Money Wage Rates in the United Kingdom, 1861-1957. Economica, 25(100), pp. 283-299.

Rudebusch, G. & Svensson, L., 1999. Policy Rules for Inflation Targeting. In: Monetary Policy Rules. Chicago: University of Chicago Press, pp. 203-262.

South African Reserve Bank, 2019. Inflation Targeting Framework. [Online]

Available at:https://www.resbank.co.za/MonetaryPolicy/DecisionMaking/Pages/default.aspx

[Accessed 7 June 2019].

Taylor, J., 1993. Discretion Versus Policy Rules in Practice. Carnegie-Rochester Conference Series on Public Policy, Volume 39, pp. 195-214.

Taylor, L., 1991. Income Distribution, Inflation and Growth: Lectures on Structuralist Macroeconomic Theory. s.l.:MIT Press.

Taylor, L., 2009. Reconstructing Macroeconomics: Structuralist Proposals and Critiques of the Mainstream. Boston: Harvard University Press