An Economist’s Advice is Only as Good as Her Data and Model

The application of economics should be seen as a craft, instead of a purely methodological process with static assumptions.

Much needs to be done to revive economic growth. Credit: WIkimedia Commons

Credit: WIkimedia Commons

“Facts are stubborn things, but statistics are pliable,” Mark Twain once said.

In a financial world as deeply integrated as today, the over reliance on growth data and metrics like gross domestic product (GDP) growth rate, gross national product (GNP) and total factor productivity (TFP) as the ultimate measures of studying a developing economy’s state of development seem a little far fetched.

In 1934, when the modern concept of GDP was developed by Simon Kuznets in a US Congress report, he mentioned how the production measurement metric should be used as a measure of gauging the increase in production capacities that is monetised in an economy and must not be used as a measure of welfare in any way; yet, today most mainstream economists, especially ‘econocrats’ (economists working as bureaucrats) consider GDP and per capita GDP as the key indicators for gauging a developing economy’s stage of development.

In the case of India, we recently witnessed how inflated data trends (explained as a rise in discrepancies in GDP growth data) contributed to a significant rise in quarterly GDP to 7.9%. In fact, data on growth accounting processes and methods focusing on aggregates need a more robust system that can transparently reflect aggregate trends with individual preferences, correcting for measurement errors and minimising discrepancies.

One of the major issues with the sophisticated growth measurement matrices remains the fundamental arbitrariness of techniques that are employed to distinguish between a production function and shifts in that function. Another issue related to growth measurement processes in emerging economies like India today is concerned with a major misspecification involved in implicitly treating growth as the sum of economic contributions made by distinct exogenous factors.

In a recent article, I pointed how growth accounting explains the difference between the contributions of increasing capital per worker and TFP – calculated  today as one of the most important production metrics. Yet, there is enough empirical research that points at a widely divergent view on the issue, oscillating between the claims that the process of capital accumulation is fundamental to a country’s economic growth, as against the notion that capital accumulation is largely unimportant in real developmental growth.

Need for robust data with countervailing measures

“The national accounts are showing, you know, a huge increase in the amount that people have, year, upon year, upon year, and we’re just not picking it up in the household surveys. That’s a very, very serious gap and I think not enough is being done to address that,” said Angus Deaton, Nobel laureate in economics in 2016. Deaton’s own work in Understanding Consumption and related studies since then, have helped highlight modelling inadequacies attached to aggregate production computation and its relationship with aggregate saving, consumption. Productivity growth and the ratio of savings to income  in growth modelling often assumes a zero rate of interest (fixed by central banks), a constant rate of growth in national income and population rates, et al.

Monetary and fiscal policies to push for more savings, however, cannot assume a zero rate of interest, which is a rare possibility as interest rates (reflecting both the cost of borrowing and rate of return) strongly affect saving capabilities amongst individuals; the link is an important one to be accounted for in shaping more effective fiscal and monetary policies. For example, if capital income is taxed more by the government, it is likely to lower real interest rates and stifle an individual’s ability to save. This can be a real response in times of recession, for instance, or the pursuit of aggressive redistributive measures as part of affirmative action by the government, which distorts savings. Thus, all these arguments depend on the existence of the positive response of saving to higher interest rates (which is assumed as zero in the growth modelling framework).

Let’s take the case of the US here: Nakamura and Steinsson discussed the pervasive problems with the import price statistics collected by the federal government. According to their research, “figures from the Bureau of Labor Statistics seemed to show that the reported price of imported furniture rose by a total of 9% from 1998 to 2007. On taking computers, official stats indicated that the price of computers for consumers fell at an average annual rate of 22% a year from 1998 to 2007, which seems to fit with personal experience. However, the import price index for computers show a drop of only 8% per year over the same time, which seems unlikely.” What is baffling to observe (according to the evidence provided by Nakamura and Steinsson) is how the import price of furniture rose over a stretch when the price paid by consumers fell by 7%. Similar problems occur for other imported consumer durables, including motor vehicles and parts.

Thus, the process of data collection in an economy’s mechanistic growth accounting process needs strong, verifiable validation, and it is important to create strong institutional feedback mechanisms on the process itself. The presence of countervailing measures (for example, industrial data scrutinised by committees of autonomous institutions) can help larger developing economies like India and China, where data collection and its validation is a major problem and where policy formulation and budget allocations take place on indicative macro-aggregate data trends.

A robust economic data collection process (with a parity in dynamic base year pricing) thus must be subjected to sufficient scrutiny for effective policy formulation in the process of growth measurement. The economies with the best functioning market-created incentives are often the ones with the best governmental systems responsible for creating strong institutional mechanisms in monitoring measurement errors in data collection (explained as discrepancies).

The need of craftsmanship in applying models

There is nothing reductionist about the practice of economic modelling till we confuse ‘a given model’ with ‘the model’ in identifying a developing economy’s path to greater economic prosperity.

Economic analysis and modelling, if skilfully crafted, can be very useful in identifying areas of potential gains, avoiding rent-seeking tendencies over time. Let me cite an example here in reference to regulating the housing market (financial deregulation, which was a major factor for the 2007 crisis in the US and the 1996 economic crisis in Thailand): while one model of the housing market (where demand and supply are perfectly competitive) shows that rent controls make housing harder to obtain, another specifies market conditions (of insufficient competition) in which modest ceilings on rents would actually increase the supply. The challenge here becomes one of choosing which model applies best to the economic situation at hand in a given economy.

The choice that requires a degree of theoretical open-mindedness and empirical investigation makes the application of economics by economists a craft, embedded in a dynamic and scientific methodological process. The practitioners of economics need to acknowledge this as an empirical fact in their policy making process, doing away with static assumptions – on rationality, institutional arrangements, political stability, time lag and the like.

For example, most of the focus in targeting for growth in countries like India is based on increasing capital investment to push for a rise in total factor productivity. However, TFP is merely a residual factor that accounts for increase in the level of production within an economy, but doesn’t necessarily explain the causes for technical changes in production levels. There are many more determinants that may lead to changing production levels, such as increased technological innovation, changes in government policies and institutional changes. It is therefore more important to look at the governing dynamics of “deeper” determinants (as economist Dani Rodrik would say) to economic growth than merely focus on productivity enhancement as an end in itself. For economists to use the methods of economics in its best scholarly capacity for a given economy, it is critical to validate the data collected for growth measurements, backed by skilful craftsmanship in applying economic principles that balance social, political aspects with local economic benefits with levels of economic and financial integration.

Deepanshu Mohan is assistant professor of economics at O.P. Jindal Global University.