Since March 2022, the US Federal Reserve has hiked its benchmark Federal Funds rate by 525 basis points (bps), from 0.25% to 5.5%, in hopes of taming inflation in the US. Given that some of that inflation was due to supply chain shocks from global lockdowns and that the US had flooded the world with liquidity since 2008, those rate hikes seem to have a rather muted impact on inflation. Economic data in the US remains robust and we therefore seem to find ourselves in a “higher for longer” world.
The rate hikes had tremendous consequences around the world, especially on exchange rates. Painfully, as we know, the ringgit has depreciated significantly against the US dollar from approximately RM4.20 in March 2022 to about RM4.70 today. To be fair, against the backdrop of rapid Fed rate hikes and the fact that the US dollar remains the global currency reserve, enjoying a so-called “exorbitant privilege”, the ringgit was just one of many emerging market countries to have depreciated against the US dollar.
However, the ringgit had a particularly dreadful performance, having the second-worst slide in Asia after the Japanese yen. But even the yen’s performance is understandable — an economy that had seen some signs of life after decades of essentially zero growth did not want to raise rates to sputter out those signs of life, thereby accepting the ongoing weakness of its currency.
Malaysia is different. Since March 2022, Bank Negara Malaysia has raised the overnight policy rate by a mere 125bps, from 1.75% to 3.00%. With Malaysian rate changes not matching those of the US, the ringgit therefore declined sharply.
A big part of the story was both growth and inflation. Raising rates too much risks jeopardising Malaysia’s economic growth post-Covid-19 movement control orders. At the same time, the nature of our economy, with heavy subsidisation of fuel, diesel and electricity, kept inflation at bay, which reduced the need for Bank Negara to hike rates. In essence, Malaysia chose domestic price controls over ringgit stability.
As a policy choice, that may not be so damaging to the ringgit if our fiscal position was far stronger. Indeed, a big reason why the ringgit was the second-worst performer in Asia is because of our current fiscal position. Until and unless we resolve this, it is difficult to imagine a long-term trajectory of both sustainable economic growth and a strengthening ringgit. And it all starts with subsidies.
The announcement by Prime Minister Datuk Seri Anwar Ibrahim on the rationalisation of diesel subsidies in Malaysia in late May was very much welcome. And as pointed out by World Bank economist Apurva Sanghi on X (formerly Twitter), this came on top of electricity subsidies rationalisation and water tariff reform. This is the right direction for Malaysia.
But much like how in video games, we start off by defeating easier bosses or mini bosses before getting to the big boss at the end (you have to sort out Saruman before you sort out Sauron, so to speak), this all sets the stage for the rationalisation of RON95 fuel subsidies. Fuelling your vehicle at a petrol station in Malaysia is cheaper than in Saudi Arabia — which makes no sense at all. In any case, as Apurva puts it, the success of the RON95 reform also requires “ensuring adequate transfers for the poor, securing middle-class buy-in and emphasising climate benefits”.
In thinking about transfers for the poor, the government’s strategy for channelling subsidy savings to cash transfers is based on data from Padu, its Central Database Hub. A day after the registration deadline on March 31, Economy Minister Rafizi Ramli said that about 17.7 million Malaysians aged 18 and above, comprising about 59% of the population, had updated their data with Padu. The government is hoping to achieve a registration rate as high as 100% before it can properly target these cash handouts.
While it is certainly good that the government is attempting to be more data-driven in its approach to economic policy, there is, at some point, a threshold of diminishing returns on how much data one actually needs relative to the costs of implementing targeted handouts. The debates on exclusion versus inclusion errors are fairly well noted; I’d probably just add that I’d rather
err on the side of mistakenly including those who should not receive handouts than on the side of mistakenly excluding those who should. And sure, with better data, we could do better targeting, but at some point, it isn’t worth the cost of implementation, which needs to be as simple as possible, for us to move forward with subsidy rationalisation.
But even within that debate, I do wonder to what extent the data can help us answer the question of targeting. And whether, from a more “meta” perspective, we may need to take a more nuanced approach to data-driven policymaking. Consider the following examples, taken from UCL professor Brian Klaas’ book, Fluke. In attempting to answer the question, “As more immigrants arrive in a country, do voters become less supportive of the social safety net?”, social scientists from Germany and the UK crowdsourced research, giving 76 research teams the same exact data and asking them to provide answers. What they found was that a quarter of the teams said yes, a quarter said no and half said, “Nothing to see here.” All from the same data set!
The second example is the Fragile Families Challenge, which traced children born to unmarried parents across 5,000 families. They collected data about the same children at ages one, three, five, nine, 15 and 22. The researchers held a competition in which they gave competing teams of scientists access to the data from the children at ages one, three, five and nine, to see who could best predict life outcomes for the children above 15 years old (recall that the original researchers had the actual data). As it turned out, all participating teams performed terribly, with even the best teams about as good as a model that just used random guessing based on simple averages.
The point of these two examples is not to say that data doesn’t matter; of course it does. Data-driven policymaking is better than one that is not based on any data at all. But there is a limit to how much data can help inform a policy decision, where the data is only as good as the biases or perspectives of the researchers in their analytical models, the analytical models themselves and the relative benefit of more data to the cost of actual ease of implementation. By all means, use data, but aim for progress, not perfection.
As such, as the government prepares for the final round of rationalising RON95 subsidies, it would do well to make “good enough” policy decisions — which is how life and evolution work anyway — rather than technically perfect policy decisions. The macroeconomic future of Malaysia depends on how quickly we can engage in sustainable subsidy reform; waiting on perfection will ultimately lead to that future being bleaker.