Counterfactual Experiments Research Articles

Quantum causal models via quantum Bayesianism

This paper presents a framework for quantum causal modeling based on the interpretation of causality as a relation between an observer's probability assignments to hypothetical or counterfactual experiments. The framework is based on the principle of ``causal sufficiency'': that it should be possible to make inferences about interventions using only the probabilities from a single ``reference experiment'' plus causal structure in the form of a directed acyclic graph. This leads to several interesting results: we find that quantum measurements deserve a special status distinct from interventions, and that a special rule is needed for making inferences about what would happen if they are not performed (``unmeasurements''). One natural candidate for this rule is found to be an equation of importance to the quantum Bayesianism interpretation of quantum mechanics. We find that the causal structure of quantum systems must have a ``layered'' structure, and that the model can naturally be made symmetric under reversal of the causal arrows.

Capital flows in the euro area and TARGET2 balances

This paper explores how the uneven recourse by national banking systems in the euro area to the ECB’s unconventional refinancing operations that led to the accumulation of large TARGET2 balances at the NCBs has contributed to the evolution of aggregate economic activity in important member states of the euro area. For the period between 2008 and 2014 we estimate a panel VAR model and identify the structural shocks by means of sign restrictions. Our results suggest that the build-up of TARGET2 balances was driven mainly by capital flow shocks while being barely responsive to other aggregate shocks. Furthermore, on the basis of counterfactual experiments we find that the ability to build up sizeable TARGET2 liabilities has contributed substantially to avoiding deeper recessions in the distressed euro area member countries like Spain, Italy, Ireland and Portugal, while to a smaller extent depressing aggregate economic activity in core member states such as Germany, the Netherlands and Finland.

To Stay or to Migrate? When Becker Meets Harris-Todaro

Allowing migration activity as an integral part of demographic transition and economic development, we establish a locational quantity-quality trade-off of children and explore its macroeconomic consequences. We construct a dynamic competitive migration equilibrium framework with rural agents heterogeneous in skills and fertility preferences. We then establish and characterize a mixed migration equilibrium where high-skilled rural agents with low fertility preferences always migrate to cities, low-skilled with high fertility preferences always stay, and only an endogenously determined fraction of high-skilled/high fertility preferences or low-skilled/low fertility preferences ends up moving. By calibrating our model to fit the data from China, we find interesting interactions between fertility and migration decisions in various counterfactual experiments with respect to changes in migration, population control and rural land entitlement policies. We conclude that overlooking the locational quantity-quality trade-off of children may lead to nonnegligible biases in assessing the implications and effectiveness of government policies.

A Shadow Rate without a Lower Bound Constraint

We propose a shadow rate that measures the expansionary (contractionary) interest rate effects of unconventional monetary policies that are present when the lower bound is not binding. Using daily yield curve data we estimate shadow rates for the US, Sweden, the euro-area and the UK, and find that they fall (rise) when market participants expect monetary policy to become more expansionary (contractionary), and price this information into the yield curve. This ability of the shadow rate to track the stance of monetary policy is identified on announcements of policy rate cuts (hikes), balance sheet expansions (contractions) and forward guidance, with shadow rates responding timely, and in line with government bond yields. We show two applications for our shadow rate. First, we decompose shadow rate responses to monetary policy announcements into conventional and unconventional monetary policy surprises, and assess the pass-through of each type of policy to exchange rates. We find that exchange rates respond more to conventional than to unconventional monetary policy. Lastly, a counterfactual experiment in a DSGE model suggests that inflation in Sweden would have been around 0.47 percentage points lower had the Riksbank not used unconventional monetary policy since February 2015.

Comparative Messaging, Learning and Forgetting: Evidence from Pharmaceutical Drug Detailing

A firm launching a new product in an existing category can leverage comparative messages to provide information to customers and to distinguish it from others. Incumbents, on the other hand, benefit less from comparative messaging as this could be interpreted as providing legitimacy to the entrant. In our empirical context in a pharmaceutical category, we see both incumbents and entrants employing comparative messages during their detailing visits. To incorporate the effects of comparative messaging, we develop a learning model of doctor prescription behavior. The model has three distinguishing features relative to the standard Bayesian learning framework. First, the information efficiency associated with the detailing visits differ across message formats (comparative vs. non-comparative) and brands. Second, comparative messages from competing brands can “jam” or disrupt a doctor’s learning of a drug’s quality. Finally, the doctors can “forget” the quality of a drug with the time elapsed from the last prescription. We estimate the model using a physician panel in a therapeutic category with one existing and two new drugs. We estimate the proposed model in a Hierarchical Bayesian framework. Counterfactual experiment characterizes the impact of banning comparative messages, a practice followed in many countries other than the U.S.

Monetary Policy and Sovereign Risk in Emerging Economies (Nk-Default)

This paper develops a New Keynesian model with sovereign debt and default. We focus on domestic interest rules governing monetary policy and external foreign currency government debt that is defaultable. Monetary policy and default risk interact as they both impact domestic consumption and production. We find that default risk generates monetary frictions, which amplify the monetary response to shocks. Large sovereign default risk depresses domestic consumption and production. These monetary frictions in turn discipline sovereign borrowing, resulting in slower debt accumulation and lower spreads. Our framework replicates the positive co-movements of sovereign spreads with domestic nominal rates and inflation, a salient feature of emerging markets data, and can rationalize the experience of Brazil during the 2015 downturn, with high inflation, nominal rates, and sovereign spreads. A counterfactual experiment shows that, by raising the domestic rate, the Brazilian central bank not only reduced inflation but also alleviated the debt crisis.

Fiscal and Individual Rates of Return to University Education with and without Graduation

Based on a detailed model of the German tax-benefit system, this paper simulates private and fiscal returns to education for college graduates and college dropouts. Completing a five-year college degree is found to be associated with an internal rate of return (IRR) of 14.2% for gross earnings, 7.4% for disposable income, and 6.6% for the net fiscal contribution. Individuals who drop out of college after two years, and subsequently complete a three-year period of vocational training, are found to have negative IRRs: -0.5% for gross earnings and -5.9% for both disposable income and the net fiscal contribution. In a series of counterfactual experiments, we explore how these returns react to changes in gross earnings, expenditure per student, and the level of income tax payments.

Transition to the property tax in China: A dynamic general equilibrium analysis

We use the dynamic general equilibrium approach to examine the potential impacts of the introduction of alternative tax regimes in the housing market in China. This approach models housing demand and supply based on the dynamic optimization behaviors of households and a representative housing developer. The model is able to capture key features in the Chinese housing market, and it provides predictions about policy effects through counterfactual experiments. We consider a universal property tax, a selective tax that applies to investment properties only, and a land value tax. Regarding the use of tax revenue, we examine two alternatives: (1) a “redistributive” scenario where tax revenue is redistributed in equal amounts to households, and (2) a “non-redistributive” scenario where tax revenue is spent on local public goods without affecting household decisions. Quantitatively we find that the selective and re-distributive tax is more efficient in terms of generating sustainable government revenues and reducing vacancy rates.

A counterfactual experiment about the eradication of cattle diseases on beef trade

AbstractIn response to disease outbreak alerts in exporting countries, importing countries usually impose trade bans that vary in terms of product coverage and in terms of duration. We rely on a unique balanced panel dataset that covers four‐digit disaggregated beef products over the 1996–2013 period, to estimate the effect of a hypothetical removal of animal diseases outbreaks on trade flows. More specifically, we investigate how bovine spongiform encephalopathy (BSE) and the foot and mouth diseases (FMDs) affect beef trade flows. We use a sectoral structural gravity approach to measure direct, conditional, and full effects, allowing inward and outward multilateral resistance indices, expenditures, and factory‐gate prices to adjust to the eradication of animal diseases. The indirect channels through which BSE and FMD impact trade are important. Our counterfactual experiment suggests that Canada would secure substantial gains from BSE and FMD eradication.

The nonlinear effects of uncertainty shocks

Abstract We consider the effects of uncertainty shocks in a nonlinear VAR that allows uncertainty to have amplification effects. When uncertainty is relatively low, fluctuations in uncertainty have small, linear effects. In periods of high uncertainty, the effect of a further increase in uncertainty is magnified. We find that uncertainty shocks in this environment have a more pronounced effect on real economic variables. We also conduct counterfactual experiments to determine the channels through which uncertainty acts. Uncertainty propagates through both the household consumption channel and through businesses delaying investment, providing substantial contributions to the decline in GDP observed after uncertainty shocks. Finally, we find evidence of the ability of systematic monetary policy to mitigate the adverse effects of uncertainty shocks.

Invest in Information or Wing It? A Model of Dynamic Pricing with Seller Learning

Pricing products such as used cars, houses, and artwork is often challenging, because each item is unique, and the seller, ex ante, lacks information about the demand for individual items. This paper develops a dynamic pricing model for products with significant item-specific demand uncertainty, in which a forward-looking seller learns about the item-specific demand through an initial assessment, as well as during the selling process. The model demonstrates how seller learning, through several mechanisms, can lead to the commonly observed downward trend in the prices of individual items. These mechanisms include the seller’s optimal adjustment of prices over time to account for the dynamic adverse selection of unsold items and the diminishing option value in future learning. The model is estimated using novel panel data of a leading used-car dealership. Counterfactual experiments show that the value of learning in the selling process is $203 per car. Conditional on subsequent learning in the selling process, the initial assessment further improves profit per car by $139. With the dealer’s net profit per car being about $1,150, these estimates suggest a potentially high return to taking an information-based approach toward pricing products with item-specific demand uncertainty. This paper was accepted by Juanjuan Zhang, marketing.

FRICTIONS AND MISMATCHES IN THE LABOR MARKET

We develop an infinite-horizon dynamic search model to understand education–job mismatches in the labor markets where job seekers face three different types of labor markets based on their minimum educational requirements. Using a new dataset, we find that our model matches the US data well when we introduce heterogeneity through wage distributions. We use counterfactual experiments to show that even when the general unemployment level is kept constant, if the conditions within different job market types change, overeducation levels may increase or decrease dramatically. We find that regardless of the general unemployment level, frictions in the job market is the main reason for overeducation. However when unemployment is high, highly educated job seekers may settle for jobs below their education level at a higher level leading to a high degree of overeducation in the labor market and crowding out job seekers who have lower level of education.

Competition and Strategic Incentives in the Market for Credit Ratings: Empirics of the Financial Crisis of 2007

We study the market for ratings agencies in the commercial mortgage backed securities sector leading up to and including the financial crisis of 2007–2008. Using a structural model adapted from the auctions literature, we characterize the incentives of ratings agencies to distort ratings in favor of issuers. We find an important equilibrium distortion, which decreased after the crisis. We study several counterfactual experiments motivated by recent policymaking in this industry. (JEL D44, G01, G24)

The marriage gap: Optimal aging and death in partnerships

In this paper we explore how marriage affects longevity of men and women through income pooling and public-goods sharing as well as joint utility maximization of partners with different preferences and biology. We integrate joint decision making of couples into a biologically founded life-cycle model of health deficit accumulation and endogenous longevity, calibrate the model with U.S. data, and perform the counterfactual experiment of preventing the partnership. We elaborate four economic channels and find that, as singles, men live 8.5 months shorter and women 6 months longer. We conclude that about 25% of the marriage gain in longevity of men can be motivated by economic calculus while the marriage gain for women observed in the data is attributed to selection or other (non-standard economic) motives.

Merger effects with product complementarity: Evidence from Colombia’s telecommunications

Mergers of firms producing complementary products entail two opposing effects: Lower prices, because they may internalize the impact of the complementarity, and higher prices, because the firms gain the ability to price discriminate. I use Colombian data to analyze mergers between firms providing complementary telecommunication services. I estimate a discrete choice model of demand for bundled fixed-line and mobile internet services, in which the degree of either substitutability or complementarity among products is a parameter of interest. Counterfactual experiments using the estimated model indicate pro-competitive effects of mergers with complements: despite a small increase in the price of standalone goods, consumer surplus increases.

Economic fluctuations, volatility changes and the role of government spending in China: A structural analysis

AbstractWe study the economic fluctuations in China by using a standard neoclassical general equilibrium model to provide a structural analysis. We have carefully constructed measurements for economic variables from Chinese data to be consistent with the literature. We show that the government spending behaviour plays an important role in accounting for the changes in the pattern of both absolute and relative volatilities. Although we find that a general moderation in economic fluctuations after 1978 can be largely explained by the total factor productivity (TFP) process, TFP itself cannot explain the change in the pattern of relative volatilities. We show that policy changes in government spending can account for the relative volatility divergency. Counterfactual experiments are also provided to discover the role of each factor in explaining the economic fluctuations in China.

Accounting for labor gaps

This study explains the impact of taxes and labor market institutions on the total hours observed in France, Germany, the United Kingdom, and the United States. We develop a balanced growth model with matching frictions in the labor market distinguishing between the extensive margin and intensive margin of labor supply. We show that (i) hours are more sensitive to changes in taxes, whereas employment reacts more to shifts in labor market institutions and (ii) a substitution effect exists between employment and hours. Counterfactual experiments show that if France had experienced the same trend in labor market institutions as the United States, its employment rate would have increased by 25 percentage points, whereas its number of hours worked per employee would have reduced by 1 percentage point. If France had chosen the US’ paths of both taxes and labor market institutions, then its employment rate would have been larger by 20 percentage points and the number of hours worked by employees would have been larger by 3 percentage points than the current one, a situation observed before the 1970s.

SNAP and Food Consumption among the Elderly: A Collective Household Approach with Homescan Data

How should we evaluate transfer programs when the targeted households are composed of individuals with heterogeneous preferences, asymmetric bargaining power, and disparate incentives to share public goods? This paper estimates a collective household model for evaluating the Supplemental Nutrition Assistance Program (SNAP) among older households. I use longitudinal Homescan data to identify SNAP-eligible food. I find that husbands have relatively stronger preferences for food than wives, and that household demand is affected by bargaining power (i.e., control over resources) within households. Failure to account for this difference in preferences and control leads to underestimates of elderly couples' total food demand, and of their implied response (at both intensive and extensive margins) to a counterfactual experiment of replacing SNAP with a cash transfer program. I find that for most eligible older households, their SNAP-eligible spending given the cash transfer is above the program's needs standard. Their spending patterns suggest that low income is the main reason for food insecurity, not that they have different preferences from richer households. Overall these findings imply that grocery vouchers or cash transfers would be more cost-effective than in-kind transfers among older households.

Bidding for Contracts under Uncertain Demand: Skewed Bidding and Risk Sharing

Procurement projects often involve substantial uncertainty in inputs at the time of contracting. Whether the procurer or contractor assumes such risk depends on the specific contractual agreement. Using auction data from the Florida Department of Transportation, we document evidence of i) risk-balancing behavior through the formation of bid portfolios, and ii) opportunistic behavior via skewed bidding. We develop and estimate a model of bidding for contracts where bidders have multidimensional private information. In equilibrium, bidders balance skewed bidding and risk exposure; both efficient and inefficient bidders bid aggressively via skewed bidding. Counterfactual experiments suggest that the onus of bearing project risk should fall on the procurer (contractor) when project risk is large (small).

A Dynamic Structural Model of Endogenous Consumer Reviews in Durable Goods Markets

Consumers commonly seek quality information from product reviews when their perceived quality of new products remains uncertain. However, consumer-generated product reviews are endogenous and may be subject to potential biases that impact consumer's purchase behavior. In this paper, We develop a dynamic structural learning model and a consumer review-reporting model to quantify the causal impacts of critic reviews and consumer reviews biases on forward-looking consumer adoption behavior in the U.S. video-game market. Our datasets consist of a novel component of pre-order sales (sales realized prior to product launch), post-release sales and prices at product level from 2009 to 2012, and individual-level online critics and consumer ratings. Our study includes two parts. First, we leverage pre-order information to separately identify the impacts of critics and consumer's prior expectation on purchase behavior. Second, we find systematic differences between critics and consumer ratings: the average of consumer ratings tends to become lower (higher) when critics are higher (lower). Therefore, a standard Bayesian learning model cannot be directly applied since such systematic differences are not merely driven by sampling errors when quality signals are drawn from distributions of true product quality. We propose that the observed differences could be driven by consumer review biases due to a combination of self-selection effect and reference point effect. We propose a novel review-reporting model that rationalizes the observed discrepancies between critics and consumer ratings. We then implement a joint estimation strategy to estimate the combined model using Bayesian Markov chain Monte Carlo (MCMC) algorithm that is applicable to a non-stationary dynamic discrete choice model with potential price, and consumer review endogeneity problems. Using structural model estimates, we examine the causal impacts of critics and consumer reviews on firms' profits. To assess these impacts, we conduct counterfactual experiments by sequentially eliminating critics and consumer reviews. The results of the proposed model can provide important managerial implications with respect to dynamic pricing and preorder strategies.