SUBSCRIBE TO THE BULLETIN TODAY!

Polishing the Crystal Ball

Past financial secretaries have overshot their last five budgets by an average of 23.7 percent. DAVID O'REAR asks if Mr Tang can get closer to the mark

When you find that every single day your car won't start, you do something about it. It wouldn't make sense to assume that just because it didn't start today, it will work just fine tomorrow and the next day and the day after that. Instead -- if you can afford the repair bill -- you get it fixed. If not, you take the bus.

That's forecasting, and it works best if you make cautious assumptions about what will change. When gazing into the crystal ball, it usually pays not to have a certain outcome in mind. For the stock analyst, putting a smiling face on the outlook might encourage more investment, thereby raising commissions and the year-end bonus. When it comes to government budgets, the temptations are just as great.

Every year in March the Financial Secretary reveals his forecast for the next five years, and on that basis, formulates policy. Of course, the forecast is updated periodically, but once a change in taxes or spending has been set in place, it is difficult (and a bit embarrassing) to turn around and admit that the assumptions were wrong.

When the Financial Secretary, back in March 2000, looked forward to the 2003-04 fiscal year, he thought we would have a better economy, and based his forecast spending and revenues, and the resulting budget balance and reserves on that assumption. (Economic forecasts are made on a calendar year basis, while budgeting follows the fiscal year, which begins April 1). The results, shown in the table below, turned out to be less favourable.

Were these errors unusual, the product of special circumstances such as terrorism and SARS? Unfortunately, no. On average, the past five forecasts, for the period up to the end of 2003, envisaged 5.3 percent growth in nominal GDP and 0.9 percent inflation, rather than the -1 percent GDP contraction and -3.1 percent deflation that we actually experienced. To put it in perspective, the current 3.5 percent medium-term economic outlook would, if the pattern holds, probably end up being between 2.7 percent and 3.1 percent.

Budgets have to look ahead, further than a year, to plan for both income and expenditure. But, when the size of the economy has not successfully been predicted to within 20 percent, problems arise. Revenues are over-estimated and the demand for services such as welfare and education (for those leaving the poor labour market to return to school) are under-estimated. The result was a multi-year deficit that cut our fiscal reserves in half.

Compounding errors

Over five budgets, the average error in estimating government revenue was 23.7 percent, with a range of 4.5 percent to 40.6 percent, all of them high. This, in turn, led to budgeting for higher expenditure that turned out to be difficult to finance from recurrent revenues. Forecasts for spending were 7.6 percent higher, on average, ranging from a forgivable 3.4 percent overshoot to an extra 12.6 percent.

Something had to give, and in our case it was the reserves. In only one year did reserves come in higher than forecast in the previous year's budget: reserves at the end of March 2001 were expected to be $383.2 billion, but ended up at $430.3 billion. Even so, and including that sole better-than-expected outcome, the forecasts for fiscal reserves just 12 months later were still overly optimistic by between 7 percent and 27 percent.

Stretch the forecast out two years, and accuracy suffers even further. At five years, a finger in the air is among the better predictors. As anyone who has planned a corporate five-year plan knows, the fourth and fifth years are pure fantasy, and should be taken as merely suggestive of the direction of outcomes.

Be cautious, Mr Tang

The Financial Secretary cannot draw up a five-year budget without forecasting the economy on which his revenues and expenditures are based, but recent predictions have proven to be unreliable. Moreover, the margin of error is very broad, which means we can_t even guess how wrong the official forecasts might be. What should be done?

Rather than abandon forecasting, or somehow improve the output (a better crystal ball, perhaps?), it would be better to take a very conservative view, down playing economic growth and revenues and bumping spending up a notch. If the consensus is that the economy will grow 4 percent a year over five years, plan for 2 or maybe 3. If deflation is expected to end, assume it does not, and profits and salaries taxes do not improve.

Under such a forecast, the hard decisions would force themselves to the fore. Further spending increases would be clearly unsustainable, reserves would vanish into thin air and the prospect of continuing on as if nothing is wrong simply untenable. Under such a forecast, the budget-cutting scalpel would become a machete.

This might seem to be an odd, or even drastic solution, given that most forecasters have been predicting good growth this year and many are raising their estimates of just a few months ago. But the alternatives are much worse. If the deficits continue, even at only $30 billion a year or so, the reserves will dwindle to a point where currency speculators start sharpening their own bladed instruments.

The up-side is that better-than expected economic growth will enhance revenues, giving us the ability to rebuild the reserves, or at least slow the pace of deterioration.

David O'Rear is the Chamber's Chief Economist. He can be reached at david@chamber.org.hk